-
Sperm Cornpetition and Alternative Mating Tactics in Bluegill
Sunfish
Peng Fu
A thesis submitted in conformity with the requirements for the
degree of Master of Science Graduate Deparûnent of Zoology
Ontario Institute for Studies in Education of the University of
Toronto
@3 Copyrighted by Peng Fu 2000
-
National Library Bibliothèque nationale du Canada
Acquisitions and Aquisitions et Bibliographie Seivices services
bibliographiques 395 Wellington Street 395. nie Wellington Ottawa
ON K1A ON4 OttawaON KIAON4 Canada Canada
The author has granted a non- exclusive licence aliowing the
National Library of Canada to reproduce, loan, distribute or sel1
copies of this thesis in microfom, paper or electronic formats.
L'auteur a accordé une licence non exclusive permettant a la
Bibliothèque nationale du Canada de reproduire, prêter, distribuer
ou vendre des copies de cette thèse sous la forme de
microfiche/nlm, de reproduction sur papier ou sur format
électronique.
The author retains ownership of the L'auteur conserve la
propriété du copyright in this thesis. Neither the droit d'auteur
qui protège cette thèse. thesis nor substantial extracts fiom it Ni
la thèse ni des extraits substantiels may be printed or otherwise
de celle-ci ne doivent être imprimés reproduced without the
author's ou autrement reproduits sans son permission.
autorisation.
-
Sperm Cornpetition and Alternative Mating Tactics in Bluegill
Sunfish
Abstract of a thesis submitted in confonnity with the
requirements for the degree of Master of Science, Department of
Zoology
University of Toronto
Peng Fu, 2000
ABSTRACT
Sperm competition is one of the most active areas of research in
mating systems and behavioural
ecology. This thesis examines sperm cornpetition in the context
of the alternative male mating
tactics in bluegill sunfish (Lepomis macrochinïs), and contains
two chapters. Chapter one
examines the fertilization success of alternative mating tactics
under sperm competition, and
tests the hypothesis that sneaks win under sperm competition, as
predicted by the Sneak-Guard
rnodel. Using microsatellite DNA fingerprinting, statistical
models, and behavioural mesures in
the field, 1 provide the fiat confirmation of this prediction.
Chapter two examines the sperm
investment strategies and mechanisms by which sneaks win. 1
identify differences between the
male mating tactics in their investrnent in gonads, sperm
characteristics, ejaculate sperm density,
and competitiveness. Overall, this thesis provides a mesure of
fertilization success under sperm
competition and an explanation as to how this success c m be
obtained.
-
I thank my supervisor, Mart Gross, who was instrumental in my
academic development
for the past five years. 1 am extremely gratefùl for his
direction and support during both my
undergraduate and graduate studies. His passion for science and
nature has left an indelible mark
on me, and 1 will always chensh the learning expenence that I
have had both in the lab and in the
field. I also thank Joe Repka for his time, knowledge, guidance,
and most of all, stimulating
discussions. 1 thank Bryan Neff for being both an invaluable
colleague and a dear friend.
Finally, I thank Peter Abrams and Maydianne Andrade for their
time and cornrnents on this
thesis.
The work on bluegil l benefited fkom the facilities of the
Queen's University Biological
Station at Lake Opinicon, Ontario, and 1 thank Frank Phelan and
Floyd Conner for their
assistance and expertise. 1 also thank my field assistants, Anna
Lawson and Tracy Michalak, for
al1 the hard work and a fabulous field season, and Karin von
Ompteda, Cory Robertson, and
Michael Berends for their assistance with laborartory
analyses.
During the past year, 1 have had the benefît of several great
fnends. 1 thank Julian Olden,
Trevor Pitcher, and Pedro Peres for stimulating conversations,
fun-filled shenanigans, and
mernorable outings. Graduate school would not have been the same
without these guys. Finally,
I would like to thank my family for their constant support and
advice.
-
Absti-act
Acknowledgements
Table of Contents
List of Figures
List of Tables
List of Appendices
Chapter one. Who wins under sperm competitioo? The Sneak-Guard
mode1 in
bluegill sunfish.
Summary
Introduction
Calculating who wins in sperm cornpetition
Methods
Results
Discussion
References
Appendices
Chapter two. Alternative male mating tactics and sperm
investment in bluegill
sunfish
Fo n;v ard
Summary
Introduction
Methods
Results
Discussion
v
vi
vii
-
Chapter Figure Title Page
One
Two
1 A demonsiration of the importance of both behavioural and
genetic 22
data in evaluating sperm competition success.
Distributions of probabilities associated with estimates of
the
relative egg number and cuckolder fertilization successes.
Scbematics of the Monte Carlo simulation used to solve for
the
average fertilization success of different categories of males
under
sperm competition.
The relationship between testes weight and body weight in
male
bluegill sunfish.
The average spem longevity and ejaculate spcrm density of male
64
bluegill sunfish.
Tne relationship between ejaculate sperm density and testes
weight
in male bluegill sunfish.
-
Chapter Table Title Page
One 1 Summary of variables and parameters used in the maximum
28
likelihood model.
Input data for the Monte Carlo simulation analysis of the
success of 34
sneaken and satellites under sperm competition with parentals
and
paternity of parentals in their broods.
-
Chapter Appendix Title Page
One 1 Description of the Monte Carlo simulation used for the
26
determination of sperrn competition success.
Procedure for the selection of an optimal number of loci for
usage in microsatellite DNA fingerprinting of parental
broods.
Input data for the Monte Carlo simulation analysis of the
success of sneakers and satellites under sperm competition
with
parentals and paternity of parentals in their broods.
-
CWTER ONE
Who wins under sperm cornpetition? The Sneak-Guard Mode1 in
Bluegill Sunfish
-
SmmlARY
Sperrn competition is a major force of sexual selection, but its
implications for mating
system and life history evolution are only beginning to be
understood. An important model for
understanding sperm competition is the Sneak-Guard model of
Parker (1 990, 1998). The mode1
rnakes two key predictions: (1) sneaks will have greater
ejaculate expenditure than guards; and
(2) sneaks will win in sperm competition. Only the first of
these has been adequately tested.
We now test the second prediction in bluegill sunfish (Lepomis
macrochirz~s). Bluegill have
both cuckolder (sneak) and parental (guard) males. Parentals
make nests, court females, and
provide solitary parental care for the embryos. Small
cuckolders, termed 'sneaken', dart in and
out of nests to ejaculate during female egg releases (dips).
Larger cuckolders, termed 'satellites',
mimic females and ejaculate behveen the parental and his h i e
female mate. Using field
behavioural data and genetic patemity analyses, we suggest that
cuckolders fertilize more eggs
than parentals when they engage in sperm competition. This
confims the second prediction of
the Sneak-Guard model: sneaks win in spem cornpetition.
-
1. LNTRODUCTION
Sperm competition is widespread in nature (Birkhead & Mdler
1998). Parker (1998)
defmes sperm competition as the "competition between the sperm
fiom two or more males for
the fertilization of a given set of ova". The ramifications of
sperm competition in the evolution
of mating systems and life histones are only beginning to be
understood. Parker (1990, 1998)
developed the "Sneak-Guard" model to undentand sperm strategies
in breeding systems where
g a r d e s (typically larger and older males) attempt to
monopolize females against sneaks
(typically smaller and younger males), which attempt to steal
fertilizations. Sneak-Guard
breeding systems are widespread among animal species and have
often evolved into alternative
male mating tactics and strategies (Gross 1996, Tabonky 1998).
The model makes two major
predictions: (1) the sneak male should have greater ejaculate
expenditure than the guard male;
and (2) the sneak male should have higher patemity per ejaculate
than the guard male. Several
studies have confirmed the first prediction by showing that
sneak males make larger investments
than guard males in testes mass and other correlates of
expenses, such as ejaculate volume,
sperm activity, and sperm longevity (e.g., Gage et al. 1995,
Taborsky 1998, Peterson & Wamer
1998, Simrnons et al. 1999). However, the second prediction has
proved dificult to test. It
requires an assessment of who wins on a per ejaculate or per
mating basis, which requires
observations of these events and specific analyses ofpatemity.
The only test of the differential
sperm competition success of sneaks and guards, performed in the
laboratory using irradiated
beetles, did not report any difference (Tomkins & Simmons
2000). We now provide a test of
the Sneak-Guard model in bluegill sdsh in the wild using genetic
markers and suggest that
sneaks do win the competition.
-
a) The Sneak-Guard model
Sperm competition is an evolutionary game between rival males
for which the solution
will be an evolutionarily stable strategy (ESS) since the
probability of winning depcnds on the
shategies played by other males in the population (reviewed in
Parker 1998). Parker (1990,
1998) recognized that many mating systems will have asymmetries
in information or spenn
competition nsk, and developed the Sneak-Guard rnodel (and the
EPC model) to provide
conceptual and predictive underpinnings. In the Sneak-Guard
model, males are either sneaks or
guards but not both. It is assumed that either a small
proportion of males, or a small proportion
of matings, involve sneaks. It is also assumed that there is a
fair rafle, without any pre-
determined advantage to each male's ejaculate (each spem counts
equally), but that one male
has more information about the probability of sperm competition
than the other. The asymmetry
in information is that guards only know the mean probability of
sperm competition (p) but not
exactly when it will occur, while sneaks always know when they
will face competition. The
asymrnetry in risk is that guards will face cornpetition in only
a portion of their matings (p c 1.0)
while sneaks will always face cornpetition (p = 1.0). Thus, the
strategy of the guard is shaped by
the mean risk while the strategy of the sneak is shaped by the
guaranteed isk. The ESS
outcome is that (Parker 1998): (1) the sneak, because of his
perfect information about sperm
competition, will invest more in competition than the guard; and
(2) the sneak, assuming equal
costs of sperm, will obtain higher patemity under the matings
where sperm competition occurs
because of the greater investment.
b) Maüng system and sperm competition in bluegiil sunfish
-
Alternative male reproductive tactics and sperm competition are
very comrnon in fishes
(Gross 1984; Taborsky l998), and many species match the
behavioural assumptions of the
Sneak-Guard model. A case in point is the mating system of
bluegill sunfish (Lepomis
macrochinis) (Gross 1982, 1991). "Parental" (guard) males breed
from ages seven to ten years
and compete in densely packed colonies for nesting sites. They
attract and spawn with multiple
females sequentially and provide sole parental care to the
developing embryos in their nest.
Females release eggs in distinctive actions called "dips", and
they dip hundreds of times in a nest
with each dip containing a batch of eggs. Precociously maturing
"cuckolder" (sneak) males
employ two alternative mating tactics termed "sneaker" and
"satellite". Sneakers fint mature at
age two years and steal fertilizations From parentals by hiding
near the edge of nests and darting
in and out to get beneath the spawning pair where they release
sperm dunng female dips. Once
sneakers reach age four, they switch ro the satellite tactic and
rnirnic females to hold a temporary
position in the nest, directly between the parental and female,
during multiple female dips. Thus,
whiie sneakers have a somewhat disadvantaged position during
sperm competition relative to
both parentals anci satellites, satellites have an advantaged
position. Parental males actively
chase sneakers and aiso satellites when these are detected. The
alternative life histories of
cuckolden and parentals are part of a single conditional
strategy with both genetic and
environmental influences on the decision process detemining
which life history trajectory to
follow (Gross 1996; Gross & Repka 1998).
Sperm competition only occurs when a cuckolder (sneaker or
satellite) successfully
inmides into a dip. Less than 20% of dips (Gross, long term data
in Lake Opinicon, Ontario) are
accompanied by successful cuckolder intrusions, and therefore
matings by parentals are typically
-
without sperm competition @ < 0.20). In contrast, mating by
cuckolders is alrnost always with
sperm competition @ = 1) (Gross 1982, 199 1).
In this paper, we set out to test the Sneak-Guard model by
determining the patemity of
sneakers, satellites, and parentals under sperm competition. The
first prediction of the model,
that cuckolders will invest more in sperm production, is already
known through their relatively
larger testes and greater ejaculate sperm density (Gross 1982;
Chapter 2). The second prediction,
that cuckolders will have higher success, has not previously
been tested because such tests are
extremely dificult to conduct methodologically. A single
bluegill nest has many thousands of
embryos resulting From the spawning of multiple fernales and
males including the parental and
several cuckolden. Although Our lab has previously determined,
using molecular markers, that
the overall patemity of parentals in nests ranges ffom 40-100%
(Philipp & Gross 1994; Neff
200 l), determining the success of an individual cuckolder per
dip was an outstanding empirical
and theoretical challenge.
-
2. CALCULATING WFIO WINS IN SPERM COMPETITION
Many challenges exist in evaluating the fertilization success of
males in a mating system
like that of the bluegill. First, the success of cuckolders
under sperm competition cannot be
readily evaluated from the overall patemity in a parental's
brood since they only participate in a
fraction of the matings (Figure 1). Second, offspring cannot be
assigned to specific dips since
thousands occur and eggs cannot be collected from dips without
disturbing spawning. Third,
only the genotype of a single male, the parental, is available
for parentage analyses as cuckolders
and females disperse after spawning. Fourth, the large number of
cuckolden and females make
it dificult to use conventional methods of parentage analysis
such as exclusion (e.g.,
Chalcraborty et al. 1988) to detemine even simple estimates such
as the total success of
cuckolden in a nest (Neff et al. 2000u, b). Fifth, the large
number of putative parents erodes the
resolving power orgenetic markers since a parental male is
likely to share allrles with females as
well as cuckolders (Neff et al. 20004 b).
We now recognize that the calculation of who wins in spenn
competition will generally
require both behavioural and genetic data (Figure 1). The
behavioural data provide the
frequency at which sperm competition occun, the genetic data
provide the overall outcorne, and
a statistical mode1 will link the behavioural data to the
success achieved per mating to arrive at
competitive success. A useful statistical procedure is a Monte
Carlo simulation (see Manly
1997) that can be used to solve for maximum likelihood solutions
given the challenges descnbed
above such as the incomplete sarnpling of candidate parents and
uncertainties in genetic
offspring assignments (e.g., Neff 200 1). We have therefore
developed a Monte Car10 simulation
to calculate the competitive success of each of sneakers,
satellites and parentals in bluegill.
-
The Monte Car10 simulation generated a distribution of the
probability of observing the
proportion of offspring that were compatible with the putative
parental male genetically (i.e.,
share at least one allele at each loci genotyped) given the
cuckolder intrusion frequency and al1
possible sneaker and satellite successes in each dip under sperm
cornpetition with parentals
(Appendk 1). The most likely sneaker and satellite successes
were chosen as ones that had the
highest probability value. The 95% confidence interval for the
sneaker success estimate was
then calculated as the value that cut-off the lower and upper
2.5% of the area under the
probability distribution given the most likely satellite
estimate. The 95% confidence interval for
the rnimic success estimate was calculated analogously while
holding constant the most likely
sneaker success value. Both the success and confidence
calculation assumed that the a priori
probability distribution of the observed intrusion frequencies
is uniform, and is the least biaçed in
the absence of additional information (for an analogous
discussion, see Neff 2000, Neff et al.
20000, b).
-
3. METHODS
(a) Behavioural data
Behavioural data were collected by divers at natural bluegill
colonies in Lake Opinicon
and by observers at an experimental pool facility near the shore
during the breeding season of
1999 (late May - early July; near Chaffey's Lock, Ontario,
Canada). The pools were stocked at
the begiming of the breeding season with mature cuckolders,
parentals, and fernales from Lake
Opinicon at densities sirnilar to the lake (Gross 1982, 199
1).
Three to four divers hovered over each nest and collected data
on female dips and the
types of males that were in the nest during the dip. Satellites
were distinguished from females
since they are smaller in size and cannot accuntely emulate the
dipping motion associated with
female egg releases (Gross 1982). Nests with females were
nndomly chosen for data collection
and observed continuously until spawning ceased. Due to the
constraint on the ability of the
observers to watch more than one nest at a tirne while
accurately recording data, not al1 spawning
within a nest was recorded. However, every effort was made to
record as much behavioural
observations as possible from single nests. Just before the fky
dispersed some seven days later,
the parental and his brood were collected using SCUBA. The f ry
and adult tissue samples were
presented in 95% ethanol for genetic analyses. The same
procedure was followed in the pool
facility, except that al1 mating behaviours were recorded, by
either direct observation or videos
taken fiom elevated platfoms above the water.
(b) Genetic data
Analyses were carried out on 28 nests including 5 fiom the
pools. Three criteria were
used to select these nests from a total of 44: (1) a minimum of
50 dips was observed; (2)
-
cuckolder intrusions were observed; and (3) the parentals raised
the offspring to the dispersal
stage, at which point they were collected. Each parental male
was genotyped at 1 1 microsatellite
loci using the methods of Neff et al. ( 2 0 0 0 ~ ) . Following
Neff et al. (2000a), the parental's allele
frequencies within the breeding population were used to
determine the optimal number of loci to
be used in genotyping 46 fry randomly chosen from each brood
(details in Appendix 2). In total,
5700 genotypes were generated for 13 16 individuals.
Each parental's patemity was calculated using the Two-Sex
Patemity model, which
allows for the contribution of multiple fathen and mothen in a
single brood (Neff et al. 2000a).
Statistical confidence (95% confidence interval) in each
patemity estimate was calculated with
the Monte Car10 simulation presented in Neff (200 l) , which
approximated the Two-Sex
Confidence mode1 (Neff et al. 2000b). The Monte-Carlo simulation
was used to expedite the
calcuIations presented in the confidence model, which would bave
been too computationally
intense to solve for today's computen given the large number of
loci used.
(c) Simdafion anaiysis
A feature of the simulation model is that there is a trade-off
between the number of eggs
released by fernales in matings with sperm competition and that
this influences the success of
cuckolders. In a brood of a fixed size, the more eggs released
in rnatings with sperm competition
relative to that without, the lower the cuckolder success value.
Therefore, without knowing a
prion' the value of the relative number of eggs released under
spem cornpetition (Er, assuming
females do not discriminate within cuckolden - E, = Em), we fint
calculated the possible range
of Ec by nmning the simulation model while assuming that the
fertilization success of cuckolders
in a single dip under competition is either 0.5 or 1.0. The
solution of the value of Ec in each
-
simulation is the one that has the highest probability of
occurrence. This calculates the threshold
Ec values for which cuckolders win under sperm competition. A
biologically reasonable Ec value
was then chosen as a parameter in the mode1 From which the
patemity of parentals, sneakers and
satellites under sperm competition was estimated. In al1
simulations, the relative s d v o r s h i p of
cuckolder offspring, S, or Sm, was set at 1.6, as they are 60%
more likely to survive to become
mature Fry than offspnng From parentals (Neff2000; Neff &
Gross in prep).
-
4. RESULTS
(a) Behaviorrrai &ta
A total of 8625 female dips were observed in 20 houn and 18
minutes of active spawning
in 44 nests fiom 7 colonies in both the lake and pools. On
average, 10.3% of female dips
included sperm cornpetit ion, where both cuckolders and
parentals spawned over the released
eggs. Sneakers participated in 8.4% and satellites in 1.9% of
dips. The lake and pool
environment did not differ significantly in inmision frequencies
(Me: average of 9.5%; sneakers
6.3%, satellites 3.2%; pool: average of 6.2%; sneakers 6.0%,
satellites 0.2%; Kniskal-Wallis test:
sneaker - 2 = 1.28, d.f. = 1, p = 0.26; satellite - 2 = 2.08,
d.f. = 1, p = 0.15). The behavioural data from the 25 nests
genotyped are in Appendix 3.
(b) Geneîic data
The average patemity of parentals and cuckolders per brood is
0.8 1 k 0.15 (SD) and 0.19
2 0.15 (SD) respectively. Each patemity estimate had high
precision as a result of a narrow
confidence interval (Appendix 2). The patemities are not
significantly different in the lake and
pools (lake: 0.83 + 0.16 (SD) and 0.1 7 f 0.16 (SD); pools: 0.73
2 0.10 (SD) and 0.27 + 0.10 (SD); t-test: t = 1.3 1, d.f. = 26, p =
0.20). The genetic data, including compatible offspnng,
parental allele frequencies, and patemities and confidence
values, are also in Appendix 3.
(c) Simrilation anaipis
If both sneaken and satellites have a competitive success of 1
.O, then fernales would have
to release 1.4 (95% CI: 1.2 - I .7) times as many eggs in dips
with sperm cornpetition than
without (Figure 2a). If both sneakers and satellites have a
competitive success of 0.5, then
-
females would have to release 4.3 (95% CI: 3.3 - 5.2) times as
many eggs in dips with sperm
competition than without (Figure 2a). Therefore, when Ec = 2 -
4, cuckolders win over parentals
in sperm competition. When Ec > 4, parentals win over
cuckolders in sperm competition.
However, given that an average of 10 - 12 eggs rnay be released
by females in a dip alone with
parentals (Gross, unpublished data), it rnay be unlikely that
females cm physically release more
than 4 tirnes as many eggs (-40) in dips accompanied by
cuckolder intrusions. This large
quantity of eggs released in a single dip, which rnay be
conspicuous to observes, has not been
previously documented or observed. Therefore, it rnay be much
more reasonable to assume that
females can release a maximum of about 20 eggs per dip (Ec = 2),
and thus cuckolders probably
win over parentals in sperm cornpetition.
Assuming the Ec = 2, cuckolders on average fertilize 76% of the
eggs in a dip and
parentals fertilize the remaining 24%. More specifically,
sneakers fertilize 89% of the eggs in a
dip (95% CI: 0.74-0.98; Figure Zb), which is significantly more
than the 64% (95% CI: 0.42-
0.83; Figure 2b) obtained by satellites (t-test: t = 22.9, d.f.
= 26,p c 0.001). Even though the
satellite success estimate rnay be largely dependent on three
nests with intrusion fkequencies
above 0.1, enough resolving power exists to detect the
significant difference in competitive
success between sneaken and satellites. Further, since
satellites are outnumbered three times by
sneakers (Gross 1982), it is expect that fewer nests would be
intnided by satellites. nius, the
few parentals cuckolded by satellites rnay not be atypical, and
within cuckolders, sneakers rnay
be more successfùl than satellites in competing against
parentals on a per dip bais.
Finally, as the simulation mode1 presented has not been
thoroughly tested with known
data sets, and the true value of the relative egg aumber is
still unlaiown (E,), the conclusions are
bea interpreted qualitatively rather than quantitatively.
Additional support for the validity of the
-
simulation results is provided by the biological knowledge of
the mating system and the sperm
investment patterns of sneakers, satellites, and parentals (see
Discussion).
-
S. DISCUSSION
We have now confirmed the second prediction of Parker's (1 990,
1998) Sneak-Guard
model. Cuckolders, both sneakers and satellites, may be superior
to parentals in sperm
competition. A cornpanion paper (Chapter 2) provides detailed
biological information on the
sperm and ejaculate investment strategies of alternative tactics
and shows that cuckoldea invest
more in total spermatogenesis and ejaculate sperm density,
supporting the first prediction of the
Sneak-Guard model. Thus, the predictions of Parker's theoretical
model have empirical support.
The only other study testing the outcome of Sneak-Guard sperm
competition is the recent
investigation by Tomkins & Simmons (2000) in internally
fertilizing Onthophagus beetles that
have alternative male reproductive tactics. The dimorphic males
include fighters (major males -
hamed) and sneakers (minor males - homless). In the species
where the Sneak-Guard mode1
applied (0. binodis), they confirmed that sneakers invest more
in sperm competition than
fighten (guards). However, they did not find suppon for the
prediction that sneakers would have
greater ferti k a t ion success than guards. The y acknow Iedged
the arti ficial nature of the
experiment (only two matings per female, irradiation to measure
patemity, laboratory setting)
and encouraged hrther studies using molecular marken and natural
conditions. We have done
so here.
An assumption of the Sneak-Guard model is that sperm competition
is a 'fair raffle',
namely that a male's fertilization success is proportional to
his relative contribution to the spem
pool competing for a batch of eggs (Parker 1990, 1998).
Elsewhere we show that cuckolden
and parentals have a fair raffle when their milt is given equal
proximity to eggs, as their
fertilization success was directly proportional to the relative
number of sperm in ejaculates of
equal volume (Chapter 2). However, there are interesting
positional differences in the mating of
-
sneakers and satellites in competition with parentals (and
probably in most species with
competing males) that results in "loaded raffles", where some
males are 'favoured' and others
'disfavoured' . Parker ( 1990, 1998) predicts that males who are
always disfavoured will
compensate by producing more sperm per ejaculate. In bluegill,
sneaken are disfavoured by
their position below the parental and female, as the female dips
toward the parental and their
urogenital pores corne in close contact. In contrast, satellites
are favoured since they ejaculate
between the female and parental, as the female dips her
urogenital pore toward the satellite.
Despite this positional advantage, satellites are infenor sperm
competitors to sneakers.
We have attributed this contradiction to the greater number of
mating opportunities that
satellites obtain (Chapter 2). Even though satellites may
fertilize fewer eggs per mating, the
transition fiom the sneaker tactic to the satellite tactic is
associated with obtaining higher
reproductive success (Chapter 2). Thus, the suprnor body
position of satellites may require them
to release fewer sperm under competition but allow them to
allocate their ejaculate expenditure
in more matings. Sneaken invest more in sperrn production by
having larger GSIs
(gonadosomatic index) than satellites since they are in a
disadvantaged mating position (Parker
1990, 1998). Their lesser number of mating opportunities rnay
require them to release more
spenn under competition since their reproductive success is more
dependent on out-competing
parentals. In contrast to cuckoldea, parentals have a low risk
of sperm competition (p < 0.20)
and an advantaged position except in the rare cases when
satellites intrude (2% of dips).
Therefore they invest the least in sperm production as measured
by GSI (Chapter 2). Since they
also fertilize 4-7 tirnes more eggs than both sneaken and
satellites (Gross & Charnov 1980,
Phîlipp & Gross 1994, Neff 200 1, this paper), they are the
poorest sperm competitors because
they may budget the least arnount of spem per mating.
-
We have discovered that more eggs are release by females in dips
with cuckolder
intrusions. The effect of this on the overall patemity of
cuckolders in a parental brood is
equivalent to the scenario that cuckolders may be fertilizing
unfertilized eggs released by
females in previous dips alone with parentals. If this is me ,
then cuckolders are avoiding sperm
cornpetition by fertilizing eggs abandoned by parentals, and the
fertilization success of
cuckolders in dips under direct competition with parentals would
decrease as a result. However,
this scenario is unlikely to be tme. First, the magnitude of
sperm limitation in mating systems
such as bluegill, where males and females are in close proximity
during mating, is very small
(reviewed in Yund 2000). As such, parentals rnay fertilize most
of the eggs released in a female
dip and cuckolders may not be able to detect and fertilize the
few eggs that may be lefi over from
previous dips. Second, cuckolder intrusions always coincide with
female dips and cuckolders
specifically aim for the location of egg release (Gross 1980,
1982). This indicates that
cuckolders target new eggs releases rather than the unfertilized
eggs from previous dips.
Therefore, it is unlikely that cuckolden are fertilizing
unfertilized eggs abandoned by parentals.
Our finding that more eggs released by females in dips with
cuckolder intrusions could either
indicate the presence of female preference or cuckolder
selection for more fecund dips. We do
not yet know which process is o c c h n g , or whether both are
occuriing, or whether sneaken and
satellites in fact both experience exactly the same number of
eggs in each dip. However, this
does provide the start to a new avenue of research in this
mating system. Finally, our model for
calculating patemities in sperm competition may have broad usage
for comparative testing of the
Sneak-Guard model in the many mating systems where females
release batches of eggs over
which males compete (e.g., insects, amphibians, fish).
-
ACKNOWLEDGEMENTS
This work was supported by grants fiom NSERC of Canada. We thank
Bryan Neff,
Anna Lawson and Tracy Michalak for help with the field work
-
REFERENCES
Birkhead, T.R. & Msller, A.P. (ed.) 1998 Spem cornpetition
and sental selection. London:
Academic Press.
Chakraborty, R., Meagher, T.R, Smouse, P.E. 1988 Parentage
analysis with genetic markers in
natural populations. 1. The expected proportion of offspring
with unambiguous paternity.
Genetics 118, 527-536.
Gage, M. J. G., Stockley, P., & Parker, G. A. 1995 Effects
of alternative rnating strategies on
characteristics of spem production in the Atlantic salmon (Saimo
safur): theoretical and
empirical investigations. Phil. Trans. R. Soc. Lond. B 350,39
1-399.
Gross, M.R. 1980 Sexual selection and the evolution of
reproductive strategies in sunfishes
(Lepomis: Centrarchidue). Ph.D. thesis, University of Utah,
Utah.
Gross, M. R. 1982 Sneakers, satellites, and parentais:
polymorphic mating strategies in North
American sunfishes. Z. Tierpsychol. 60. 1-26.
Gross, M.R. 1984 Sunfish, salmon and the evolution of
alternative reproductive strategies and
tactics in fishes. In Fish reprod,icrion: strategies and tactics
(ed. G. Poas & R. Wootten), pp.
55-75. London: Academic Press.
Gross, M.R. 199 1 Evolution of alternative reproductive
strategies: fiequency-dependent sexual
selection in male bluegill sunfish. Phil. Tram R. Soc. Lond. B
332, 59-66.
Gross, M. R. 1996 Alternative reproductive strategies and
tactics: diversity within sexes. Trend.
Ecol. Evol. f l,92-98.
Gross, M. R & Charnov, E. L. 1980 Alternative male life
histores in bluegill sunfish. Proc. Nati.
Acad. Sci. USA 77,6937-6940.
-
Gross, M.R. & Repka, J. 1998 Stability with inheritance in
the conditional strategy. J. Theor.
Biol. 192,44543.
Manly, B.F.J. 1997 Randornizution. bootstmp and Monte Carlo
methods in biology. New York:
Chapman and Hall.
Neff, B.D. 2000 Genetic markers and breeding success:
theoretical and empirical investigations
in fish. Ph.D. dissertation, University of Toronto, Toronto,
Ontario.
Neff, B.D. 200 1 Genetic patemity analysis and breeding success
in bluegill sunfish (Lepomis
macrochirus). J. Hered. In press.
Neff, B.D., Repka, J. & Gross, M.R. 2000a Parentage analysis
with incomplete sampling of
parents and O ffspring. Mol. Ecol. 9 , 5 1 5-528.
Neff, B.D., Repka, J. & Gross, M.R. 2000b Statistical
confidence in parentage analysis with
incomplrte sampling: how many loci and offspnng are needed? Mol.
Ecol. 9, 529-540.
Ne& B. D., Fu, P., & Gross, M. R. ZOOOc Microsatellite
multiplexing in fish. Trans. Am. Fish.
Suc. 129: 584-593.
Parker, G.A. 1990 Sperm compeiition games: sneaks and extra-pair
copulations. Proc. R. Soc.
Land. B 242, 127-133.
Parker, G.A. 1998 Spem competition and the evolution of
ejacalates: towards a theoiy base. In
Sperm competition and se-mal selection (ed. T.R Birkhead &
A.P. Meller), pp. 3-54. London:
Acadernic Press.
Petersen, C.W. & Warner, R.R. 1998 Sperm competition in
fishes. In Sperm competition and
seruol selection (ed. T.R. Birkhead & A.P. Mialler), pp.
435-464. London: Academic Press.
Sirnmons, L.W., Tornkin, J.L. & Hunt, l. 1999 Sperm
competition games played by dimorphic
male beetles. Proc. R. Soc. Lond. B 266: 145-150.
-
Taborsky, M. 1998 Sperm competition in fish: "bourgeois" males
and parasitic spawning.
Trend. Ecol. Evol. 13,222-227.
Tomkins, J.L. & Simmons, L.W. 2000 Sperm competition games
played by dimorphic male
beetles: fertilization gains with equal mating success. Proc. R.
Soc. Lond. B 267, 1547- 1553.
Yund, P.O. 2000 How severe is sperm limitation in natural
populations of marine fiee-spawners?
Trend. Ecol. Evol. 15, 1 0- 13.
-
Figure 1. In order to determine the 'kinner" of sperm
competition in complex rnating
systems, behavioural data that differentiate matings with and
without sperm competition and
genetic data that differentiate the relative genetic
contributions of sperm cornpetitors in a brood
are both needed. This is a figure of a hypothetical brood
resembling a nest of bluegill sunfish in
which females release eggs in multiple dips (P = parental; C =
cuckolder). (a) Genetic analysis
reveals that P fertilized 75% of al1 the eggs in the brood while
C fertilized the remaining 25%.
The paternities could result from either scenario (b) or (c). in
(b), 33% of the dips involved
sperm competition and C fertilized 75% of the eggs per dip. In
(c), 100% of the dips involved
sperm competition and C fertilized only 25% of the eggs in those
dips. Although (b) and (c) are
clearly not equivalent scenarios, either could result in the
same overall patemities of P and C in
the brood. ï'herefore, behavioural data on the frequency of
sperm competition must be known in
order to calculate who is the supenor sperm cornpetitor and test
the Sneak-Guard model.
-
Figure 1.
(a) Brood
lndividual matings
-
Figure 2. (a) The distribution of probabilities associated with
estimates of the range of
possible values of the relative egg number (Ec) for which
cuckolders win in sperm competition.
Two probabiiity distributions were calculated, one with an
assumed cuckolder competitive
success of 0.5 (m) and the other with a success of 1 .O (+). @)
The distribution of probabilities
associated with estirnates of sneaker (a) and satellite (+)
fertilization success under sperm
competition with parentals with Ec = 2.
-
Figure 2.
O 2 4 6 8 10
Relative Egg Number (E,)
Patemity under spem cornpetition
-
APPENDIX 1
The variables in our mode1 are defined in Table 1. Generally,
the simulation involves
three randomization routines that are based on a defined
parameter set and a variable set. The
parameter set includes the behavioural data and the genetic
data, and the variables set includes
quantities that the mode1 calculates (e.g., relative
cornpetitive success of each category of males).
The simulation is repeated 10,000 times for each nest from which
the most-likely success of each
male category is calculated (Figure A ln, b).
The first routine in the simulation randomizes the intrusion
kquencies based on the
observed data. This routine accounts for variance among nests in
the number of rnatings
observed and hence the precision of the intrusion frequency
estimates. The second routine
randomizes the genotypes of the cuckolden and fernales that have
genetically contnbuted to the
offspring within the nest (Neff et ni. 2000~. b). This routine
accounts for the incomplete
sampling of the putative parents in the genetic analysis. The
third routine is the rnost complex
and is described in detail (Figure A 16). Briefly, it
randornizes the genotypes of the offspnng
genetically sampled from the nest based on the results of the
previous two routines and the
variable set, and determines the proportion of these offspring
that are genetically compatible with
the parental male. This proportion is then compared to the
observed proportion obtained from
microsatellite DNA fingerprinting. The likelihood or probability
of the variable set can be
determined as the number of matches out of the 10,000 simulated.
For example, if very few of
the 10,000 offspring samples at any of the nests do not match
the observed proportion of
compatible offspring, then the given variable set is unlikely.
Conveaely, if most of the 10,000
samples at each of the nests match the observed proportions,
then the given variable set is likely.
-
A global maximum solution c m be found by considering al1
possible combinations of variable
sets (i.e., the variable space).
-
Table 1. Sumrnary of variables and parameters used in the
maximum likelihood model. Parameters are known quantities inputted
into the simulation while variables are quantities outputted from
the simulation. A quantity cm be either input or output depending
on whether or not it is known. A quantity can be neither input nor
output if it is calculated exclusively in the simulation fiom the
input.
Name De finition Inputlûutput
Number of dips obse~:ed for nest E Input Number of intrusions by
sneaken observed in nest n hput (expressed as a proportion of D.)
Number of intrusions by satellites observed in nest n Input
(expressed as a proportion of D.) Effective number of fernales
genetically contributing to Input offspnng in nest n Effective
number of males genetically contributing to offkpring Input in nest
n Number of offspring sarnpled from nest n Input Number of O
ffspring genetically compatible with putative Input father in nest
n
Survivorship of sneaken' offspnng from fertilization to E ither
sample collection Swivonhip of satellites' offspnng from
fertilization to Either sample collection Sunrivonhip of parentals'
offspring from fertilization to Either sample collection
Relative number of eggs released in presence of sneaker Either
Relative number of eggs released in presence of satellite E i ther
Relative number of eggs released in presence of parental Either
Probability that a sneaker fertilized the egg during offspnng
Neither assignment in the simulation Probability that a satellite
fertilized the egg during offspring Neither assignment in the
simulation Proportion of eggs in a single dip (E,) fertilized by a
sneaker Output during spem competition Proportion of eggs in a
single dip (E,) fertilized by a satellite Output dunng sperm
competition Proportion of eggs in a single dip (E,) fertilized by a
parental Output dirring sperm competition
-
Figure Al. Schematics of the Monte Car10 simulation used to
solve for the average cornpetitive
success of different categones of males under spem cornpetition.
(a) The simulation involves
seven steps. In step 1, the behavioural and genetic data are
inputted. Steps 3-5 involve three
randornization routines and are repeated 10,000 times for each
of the N broods and each possible
parameter set. The maximum likelihood solution is determined
from the parameter set with the
highest probability of matching the observed genetic data. (b)
Schematic of steps involved in
Step 5. For each of the C. offspring sarnpled from the brood, a
cuckolder intrusion is
probabilistically simulated based on the behavioural
observations of intrusions. If a cuckolder
does not intrude then the offspring is genetically assigned to
the parental male. If a cuckolder
partakes in the mating, then Equation A l is used to determine
if the cuckolder or parental
successfùlly fertilizes the egg. The calculation of who
fertilizes the egg using Equation A l
incorporates the intrusion frequencies of sneakers and
satellites, the relative number of eggs
released by the female, and the differential survivorship of
cuckolder and parental offspnng. If
the cuckolder is successful at fertilizing the egg then a
genotype is generated for the offspring
based on those generated in Step 4. The offspring genotype is
compared to the parental male and
if genetically compatible (offspring shares at least one allele
with the parental male at each
locus) then it is assigned to the parental male. This emulates
the error in offspnng assignments
as a result of the possible similar genetic profiles of putative
fathers. The routine is repeated for
each of the C. offspring.
-
Step 1
Step 2
Step 5
Step 6
Input Parameter Set
Setup Variable Set
Step 3
Step 4
Randomize Intrusion Frequencies - for each of D, dips randornly
generate a number between O
and 1 - randomized intrusion frequency is the proportion of
these
numbers that are < (I,'" + I,")
Randomize Cuckolder and Female Genotypes
- for each of M cuckolders and F mothers randomly generate
multiIocus genotypes based on breeding population al1eIe fiequenc
ies and assuming Hardy- Wein berg equil ibrium genotype ratios
I Randomize Offspring Genotypes
Figure Ala.
1 Caleulate Probability of Observing Genetic Data
- brised on variable set and results of steps 3 & 4 generate
Cn offspring genotypes (see Figure A 1 b)
- proportion of 10,000 samples that have k, of C,, offspring
compatible with the putative father (see Figure A 1 b)
-
Step 7 Determine Maximum Likelibood Solution
- variable set with highest probability of matchhg the observeci
number of compatible offSpring
-
Figure Alb.
Step 4
Determine if Egg Release is Accompanied by Cuckolder
Intrusion
- generate randorn number between O and 1 - compare number to
randornized cuckolder intrusion rate (from Step 3)
I
* Determine if Cuckolder or Parental
Fertilizes Egg Under Sperm Cornpetition - generate mndom number
between O and 1 - determine if number < Pr, or Pr, (depending
on
if intruding cuckolder is s n d e r or satellite); Pr is
calculated according Equation A 1 (bottom).
cuckoIder wins
parental wins
Generate Offspring Genotype
- mdomIy select cuckolder's and rnother's genotypes ( h m Step
4)
- generate offspnng genotype based on Mendefian inheritance
* Y= Increment Number of Offspring that are
Compatible with the Putative Father I
Step 6
Equation A 1 : Pat,. Ii -4 -Si Ri =
(1-1, -I,)- E;S, +r; Esn*Ssn +[ , -AT, -Ssu 1 i can either be sn
or sa.
-
APPENDIX 2.
The loci selection process for each parental, developed in
accordance with the Two-Sex Paternity
mode1 presented in Neff et al. (2000a), is as follows. First,
the expected proportion of the total
next generation individuals (NGIs) that is genetically
compatible with the putative parental by
chance (NGdud) is calculated for each locus. The lower the NGdd
value, the greater the resolving
power of that locus in excluding non-compatible offspring from
the putative parental. Second,
al1 loci were ranked based on their NGdad values for each
parental. A cumulative AGdud value
was then calculated as a multiplicant across al1 loci for each
parental. The lower the cumulative
NGddd value, the greater the resolving power of that combination
of loci in excluding non-
compatible offspring From the putative parental. A cumulative
NGUd value of 0.1 was chosen as
the threshold of acceptable resolving power for a combination of
loci. Finally. the physical
properties of each locus were exarnined for their
appropriateness to be genotyped in a multiplex
(see Neff et al. 2000~). Generally, combinations of loci with
high resolving power and non-
overlapping alleles were selected to genotype the brood of each
parental.
-
APPENDIX 3.
Exact input data for the Monte Carlo simulation analysis of the
success of sneakers and satellite
under sperm cornpetition with parentals and the paternity of
parentals in their broods. The
behavioural data are compnsed of the total number of dips
observed in a parental nest and the
cuckolder intrusion Frequencies, expressed as a proportion of
al1 dips in a nest accompanied by
sneaker and satellite intrusions. The genetic data is compnsed
of the number of offspring out of
the 46 sampled that appear to be compatible with the brood
guarding parental and the parental's
combined allele frequencies at each of the microsatellite loci
used. The combined allele
frequencies are calculated according to Neff et al. (2000a). The
patemity and confidence
analysis was conducted following Nef€ et al. (2000a, b).
-
ci-
;=. I b ci- p? O d ;=.
O ci cn c?- e O' mi' - (O m' \o' vr 00' 2 2 2 22 22 C? ? V!99.
'=? '? C O O O O O C
-
CHAPTER Two
Alternative male mating tactics and sperm investment in bluegill
sunfish
-
FORWARD
Chapter Two includes some data and analyses from my ZOO 499Y
independent research
project, Department of Zoology, University of Toronto (September
1998 - May 1999). These
are described in sections (b), (c), (4, and V) of Materials and
Methods. Sections (b) and (4
were modified andor re-analyzed for this thesis. The
Introduction and Discussion of these
sections are entirely new. Al1 other data analyses are original,
and were performed during my
residency as a M.Sc. graduate student (September 1999 - October
2000).
-
SUMMARY
Although alternative mating tactics are found in the males of
many species, little is
known about their adaptations to sperm cornpetition and the
mechanism by which they obtain
fertilization success. We now report on the sperm investment
strategies of parental, sneaker, and
satellite male bluegill sunfish (Lepontis macrochirus). We find
that different investments in
testes, sperm characteristics, ejaculate sperm density, and
competitiveness can be explained by
tactic-specific differences in fertilization behaviour,
tradeoffs behveen somatic and gonadal
investment, sperm competition risk, and total mating
opportunities. We show that there is
disruptive selection on body size in cuckolden and that sneakers
make the transition to satellites
because the latter obtain more mating opportunities and greater
reproductive success. Parentals
have absolutely larger testes than both sneakers and satellites,
while sneakers and satellites have
larger testes relative io body size. Although the relationship
behveen testes size and body size is
positive in al1 three mating tactics, they are linearl y
discontinuous and there fore allometry cannot
explain the difference in absolute testes size among parentals,
sneakers, and satellites.
Furthemore, while the size of sperm did not differ between the
three mating tactics, parentals
had longer lived sperm than both sneakers and satellites.
Finally, the ejaculate sperm density of
satellites was greater than that of sneakers, which in ni was
greater than that of parentals. We
find that the success of males under competition is at least in
part attributed to the number of
sperm released in the ejaculate and thus provide confirmation to
the theoretical assumption that
sperm cornpetition may operate as a "rafle".
-
1. INTRODUCTION
In many species, males have alternative mating tactics (Gross
1996). These males often
engage in sperm competition, with the ejaculates from males with
different tactics competing to
fertilize the same egg(s) (Parker 1998; Taborsky 1998). Parker
(1 990, 1998) proposed the
Sneak-Guard mode1 of sperm competition in which certain males,
the guarders, are paired
permanently to females while other males, the sneakers.
specialize at stealing fertilizations. He
predicted that sneaks should be supenor sperm competiton because
they typically experience
much higher levels of sperm cornpetition than the guards. In a
cornpanion paper, we provide the
first confirmcition of this prediction by demonstrating that, in
bluegill sunfish (Lepomis
macrochinrs), the sneak-like cuckolders are superior spem
competi tors to the guard-li ke
parentals (Chapter 1). We now address the mechanism by which
supenor fertilization success
can be obtained.
Sperm and ejaculate investment by males with alternative mating
tactics has been studied
in only a few systems. Gage et al. (1995, 1998) compared
precocious parr (sneakers) and
anadromous males (guarders) in Atlantic salmon (Salmo salar) and
found that parr invested more
in total spermatogenesis, which was measured by testes size and
sperm number relative to body
size in stripped ejaculates. The sperm morphology of the two
mating tactics was similar, but pan
sperm lived longer. Stockley et al. ( 1 996) f o n d that
fertilization success in the common shrew
(Sorer araneus) was higher among guarders with absolutely larger
testes and sneakers with
greater sperm counts. Taborsky (1998) suggested that the
investment in testes versus body mass
differed between sneakers and guarders in the European ocellated
wrasse (Symphodus ocellutus),
with the relationship being negative in sneakers but positive in
guarders. Schiirer & Robertson
(1999) found that, in bluehead wrasse (Tholassoma bifasciatum),
the srnaller and more
-
promiscuous initial phase (IP) males had greater ejaculate sperm
density and invested more in
daily ejaculate expenditure than the Iarger and more solitary
terminai phase (TP) males. The
sperm of these two tactics did not differ in size. Overall,
these studies suggest that the sneak
tactic typically invests more in sperm and ejaculates than the
guarder tactic, and that this
elevation in investment is typically conelated w-ith the higher
level of sperm competition
experienced by sneakers. However, as implicitly assumed in the
Sneak-Guard mode1 of Parker
( 1990, 1998), the trade-off between obtaining mating
opportunities and fertilization success
within matings rnay also effect sperm and ejaculate investments.
In this paper, we investigate
the iactic specific differences in sperm and ejaculate
investrnent in male bluegill sunfish and
address the relationship between mating opportunities. sperm
investment, and competition
success.
Bluegill mating system
Male bluegill sunfish have a discrete polymorphism in life
histones termed "parental"
and "cuckolder", and both spawn in the same nests (Gross 1982).
While parentals are
specialized nest and mate guarders, cuckolders are specialized
sperm competitors that gain entry
into parental nests as either "sneakers" or "satellites".
Sneakers are cryptic and satellites are
female mirnics. Male bluegill, therefore, are characterized by
three mating tactics (parental,
sneaker, and satellite), and sperm competition occurs when
sneakers or satellites intrude upon
matings between parentals and females. The characteristics of
the mating tactics that are
relevant to this study are: (1) parentals experience sperm
competition in less than 20% of
matings while cuckolders experience spem competition in nearly
100% of matings (Gross, long
term estimate in Lake Opinicon, Ontario); (2) parentals
fertilize 4-7 times more eggs than
-
cuckolders in a breeding season and thus have greater mating
opportunities (Gross & Chamov
1980; Philipp & Gross 1994; Neff 200 1); (3) larger
parentals obtain more mating opportunities
and eggs than smaller parentals (Gross & MacMillan 198 1);
(4) cuckolders are superior spenn
competitors than parentals shce they fertilize more eggs in a
"dip*' or rnating (Chapter 1); (5)
sneakers may be better sperm competitors than satellites
(Chapter 1); (6) satellites gain closer
proximity to fernales during spawning since they ejaculate
directly between the parental and
female while sneakers are usually undemeath (Gross 1980, 1982;
Gross, unpublished data); and
(7) sneakers and satellites do not usually compete against each
other, as sirnultaneous nest
intrusions by multiple cuckolden are rarely observed (Gross
1980; Gross, unpublished data).
Predictions
We formulate a number of predictions fiom the literature and our
earlier work regarding
the spem and ejaculate investment strategies of parental,
sneaker, and satellite male bluegill.
These predictions are empirically tested, and several are
confirmed while others are rejected. We
evaluate the significance and ments of these predictions in the
Discussion.
(a) Cuckolder mating opportunities
We predict that (i) intermediate sized cuckolders should be the
least successful in
obtaining mating opportunities. Dimptive selection (e.g., Gross
1985) should favour smaller
sneakers and larger satellites. Smaller sneakers are more
cryptic while larger satellites more
closely match fernales in size (Gross 1982).
-
We predict that (ii) parentals will have absolutely larger
testes than cuckolders, and that
(iii) parental gonad sue will positively correlate with body
size. These predictions are based on
the knowledge that parentals may optimize gonadal investment
(Shapiro et al. 1994) so as to
fertilize the 4-7 tirnes more eggs that are available to them,
and that larger parentals fertilize
more eggs than smaller ones (Gross 1982; Philipp & Gross
1994; Neff 200 1). Within
cuckolders, we predict that (iv) satellites should have
absolutely larger gonads than sneakers,
since the developmental transition fiom sneaker to satellite
implies that satellites fertilize more
eggs (Gross 1982). Gonads should (v) decrease in size with
increasing body size in sneakers and
should (vi) increase in size with body size in satellites. This
is due to the reduced nurnber of
mating oppominities for cuckolders of intermediate sizes.
We predict that (vii) cuckolders have a larger gonadal-somatic
index (GSI) than
parentals, sincc cuckolders are specialized parasites that use
their soma more purely as a vehicle
to deliver spem, while parentals use their soma also to
constnict nests, guard mates. and provide
care to the offspnng (Gross 1982; Taborsky 1998). Parker's (1
990, 1998) Sneak-Guard mode1
also predicts that cuckolden will have a larger GSI, since
cuckolders almost always mate under
spem competition while parentals only rarely mate under sperm
competition. Within
cuckolders, (viii) sneakers will have a larger GSI than
satellites since sneakers must compensate
for their relatively disadvantaged body position (Gross 1982;
Parker 1990; Gage et al. 1995).
(c) Sperm morphology
We predict that (k) cuckolders and parentals will ciiffer in
sperm morphology. It has
previous been docurnented that sperm competition favours the
production of larger sperm, which
may be faster or more cornpetitive (e.g., Radwan 1996; Otronen
et al. 1997; LaMunyon & Ward
-
1999; Simmons et al. 1999). Therefore, as cuckolders experience
more sperm competition than
parentals, cuckolder sperm should be larger. However, it should
be noted that two other shidies
in fish failed to fmd a difference (e.g., Gage et al. 1995,
1998; Schiirer & Robertson 1999).
(d) Sperm longeviv
We predict that (x) cuckolders will have longer lived sperm,
reflecting a cornpetitive
advantage of greater spem energetics (e.g., Gage et al. L 995,
1998).
(e) Sperm qiiantity
We predict that (xi) cuckolders will have greater sperm density
in their ejaculates than
parentals. Cuckolders, but not parentals, are Iimited in body
cavity for the storage of milt (Gross
1982). Furthemore, since cuckolden are almost always under spem
cornpetition, they may
require denser ejaculates to release a larger nurnber of sperm
more quickly, as suggested by
Scharer & Robertson (1999). Within cuckolders, (xii) sneaken
will have the greatest sperm
density since they are most constrained for body cavity space,
and higher densities may be
required to release sperm more quickly due to the short of
amount of time that they spend in
nests.
fl Ejacit late competitiveness
We predict that (xiii) the competitiveness of bo!h cuckolder and
parental ejaculates
should be correlated with the number of sperm released. This
relationship has been assumed
(e.g., Parker 1998) but never demonstrated.
-
2. MATERIALS AND METHODS
Behavioural observations at colonies, and CO llections of
breeding bluegill including
cuckolders, parentals, and gravid females, were made in Lake
Opinicon, Ontario dunng the
breeding seasons of 1984, 1996 and 1999 following the methods
described by Gross (1980,
1982). Data from different years of collection were not combined
in subsequent analyses.
(a) Cuckolder mating opportzmities
Mating observations were made at 36 nests in five colonies in
three spawning bouts.
Female dips were counted, and the number and types of males that
partook in spawning were
recorded. Each spawning cuckolder was visually classified as
either small or large sneaker and
small or large satellite. Small sneakers and large satellites
were assumed to be ages 2 and 5
yean respectively (Gross 1980, 1 982). Since intermediate sized
cuckolders are known to adopt
either the sneaker or satellite tactics, large sneakers and
srna11 satellites were both assumed to be
of ages 3-4 years old (Gross 1980, 1982). The ratio of the
number ofindividuals in each of the
four size category of sneakers and satellites is 1 : 0.42 : 0.40
: 0.069 (Gross 1982, Table 5). The
total intrusions Frequency, and hence the relative number of
mating opportunities, for each
cuckolder size class was caiculated as the number of successfbl
inmision divided by the total
number of dips observed, The relative intrusion frequency per
individual for each cuckolder size
category was calculated as the total intrusion fiequency divided
by the relative size of the age
cohort. Assuming equal mortality rates and a constant population
size over rime, the theoretical
average reproductive success for individuals who remain as small
sneakers throughout their
reproductive lifespan (RS,) and those who make the transition
from sneakers to satellites (Rli,)
c m be calculated as:
-
Here, FS' and FS, are the average per dip fertilizadon success
of sneakers (0.89) and satellites
(0.64) when they intrude into parental nests and engage in sperm
competition (Chapter l), and
I.,, 6 , I,,, and b are the relative, per individual intrusion
Frequencies of small sneakers, large
sneakers, srnall satellites, and large satellites respectively
.
The behavioural data were collected by P. Fu and field
assistants during the field season
of May - M y 1999.
(b) Gonadal invesmeni
Forty-eight sneakers, 27 satellites, and 69 parentals collected
at colonies before spawning
were weighed, and their testes were removed and also weighed.
The gonadosornatic index (GSI)
for each tactic was calculated as GSI = (gonad weightftotal body
weight) x 100.
The body and testes measurement data were collected by P. Fu,
B.D. NeE, and field
assistants during the field season of May - July 1996.
(c) Sperm morpltology
Milt fiom live males was obtained, preserved, and filmed using
scanning electron
microscopy (Hyat 1974; Bovola & Russell 1992). Three
sperrnatozoa images were measured
for each of 10 males (five cuckolders and five parentals) using
a digitizîng tablet and AutoCad
R12. Spermatozoa head and rnidpiece were identified following
Wicker & Huish (1982) and the
length and width of each were measured as cross-sectional
distances. The flagellurn was traced
as a hi&-resolution polyline sketch. Each sperm was measured
three times to increase precision
and the means were used in the subsequent analysis. Sperm
morphology was compared between
-
the cuckolder and parental life histories, and sperm from
sneakers and satellites were analysed
together due to the srnall sarnple size.
The spenn specimens were collected by P. Fu and B.D. Neff during
the field season of
May - July 1996, and the sperm morphology data were collected by
P. Fu in Septernber 1997 -
May 1998.
(d) Spem longevity
Equai volumes of milt were collected from 7 sneaken, 6
satellites, and 10 parentals with
graduated syringes and mounted on glass slides. The spermatozoa
were "dilution-activated"
(Scott & Baynes 1980) with lake water and observed under a
compound light microscope with
400x magnification. Sperm longevity was estimated as the time
from activation until 80% of
. the spematozoa within the field of view were no longer motile.
Measurements were repeated
four times for each male, twice by each of two observers. There
was high consistency among
observations and between the two observers (ANOVA: FJ2, = 1.65,
p = 0.2 1 ).
The sperm longevity data were collected by P. Fu and B.D. Neff
durhg the field season
of May - July 1996.
(e) Spem quantity
Ten microlitres of fiee-flowing milt were collected from 22
sneakers, 26 satellites, and 76
parentais. Two microlitres per male were diluted in 1 millilitre
of water and thoroughly mixed.
Six microlitres of the dilution were injected into a Bright-line
hemacytometer (American
Optical), and the sperm were counted in five 0.2 mm x 0.2 mm
grids under a compound light
microscope at 400~ magnification. Spem density was calculated by
multiplying the mean
-
spenn count per grid by the dilution factor and volume. The
gonads of the males were then
removed and weighed. The weight change in gonads due to the
removal of milt was negligible.
The sperm longevity data were collected by M.R. Gross and
assistants during the field
season of May - July 1984.
If) Ejacdate competitiveness
Milt and eggs were sûipped from 4 cuckolders, 4 parentais and 3
fernales. Equal
volumes of milt from a cuckolder and a parental were mixed,
dilution-activated, and the paired
milt was released over a sample of 30 - 100 eggs from each
femaie. The containers with milt
and eggs were placed in an aquarium of lake water maintained at
ambient temperature for
incubation. The embryos were reared for a week and then
preserved in ethanol. Rearing was
not successful in 6 of the 16 trials (4 cuckolders 4 parentals)
but the remaining 10 trials still
represented a thorough combination of cuckolder-parental
cornpetition pairs as eac h individual
was used in either 2 or 3 trials. DNA was isolated from the
adults and fry, and microsatellite
fingerprints were obtained using multiplex genotyping (Neff et
al. 2000). Up to five primer sets
(presented in Colbume et al. 1996; Neff et al. 1999) were
simultaneously amplified and al1
competing parental and cuckolder offspnng were unambiguously
identified using exclusion
methods (e.g., Chakraborty et al. 1988).
The experimental fertilization trials were carried out by P. Fu,
B.D. Neff and assistants
during the field season of May - July 1996, and the DNA
fingerpnnting was performed by P. Fu
in September 1997 - May 1998.
-
Statistical calculations were performed with SPSS (version 9.0).
Al1 mean values are
expressed with plus or minus one standard deviation and al1
statistical testes were performed
with two-tailed levels of significance. AI1 regression dope
cornparisons were performed using
Student's t-tests (Zar 1996) and a11 ANOVAs were perfonned
assuming an unbalanced design
with Type III sum of squares. Specifically the numbers of mating
opportunities were compared
among al1 size categories using X' tests with the expected
values calculated as the total number of
mating oppominities obtained by al1 categones divided by the
relative size of the age cohorts
(based on body size, as reported in Gross (1982)). The
relationships between gonad weight and
body weight within parentals, sneakers, satellites, and
cuckoldes as a whole were tested with
both linear and n a m l logarithmic regressions. A natural
logarithmic regression was used when
analysis of residuals from a linear regression revealed
deviations from homoscedasticity (Zar
1996). The GSI and absolute gonad weight of the life histories
were compared using an
ANOVA and Tukey's post hoc analysis. Sperm morphology was
compared between life
histories with an ANOVA with measurements of different sperms
for the same individual as
repeated measures. Sperm longevity between mating tactics was
tested using an ANOVA with
the four observations from each individual as repeated measures.
Sperm quantity between
mating tactics was tested using an ANOVA with Tukey's post hoc
analysis. The relationship
between sperm density and gonad weight within the three mating
tactics was investigated with
linear regressions. The proportional paternity data were arcsine
square root transformed to
eliminate mean-dependent variation (Zar 1996). The relationship
between absolute gonad
weight and fertilization success was examined with Spearman's
non-pararnetric statistics with 2-
tailed measures of significance on z-scored values. For example,
when testing the effect of
parental gonad weight on parental fertilization success, ranking
parental patemity and gonad
-
weight within trials with the same cuckolder using z-scores
controlled differences among
individual cuckolders.
-
3. RESULTS
(O) Cuckolder mating opportunities
In total, 7471 dips were observed in the nests of parentals.
Cuckolders attended 766
(10.3%) and the parental was alone in the remainder. The
opportunities for cuckolders varied
with their size and tactic. Small sneakers, large sneaken,
srnall satellites, and large satellites
attended 6.0% (450 dips), 1.3% (98), 1.2% (92), and 1.7% ( 1 26)
of dips, respectively. The
relative intrusion ftequency per individual cuckolder in each
category was &O%, 3.1%, 3.0%,
and 25% respectively. Therefore, on an individual basis,
satellites obtained about three times as
many matings as sneakers. Cuckolden of intermediate sizes were
the least successful at
obtaining mating opportunities while large satellites were the
most successful. If cuckolden
were to remain as small sneaken for their entire reproductive
life span (4 years), their relative
average reproductive success would be 0.053. When cuckolders
rnake the transition from
sneakers to satellites, their relative average reproductive
success would be 0.065. Therefore,
sneakers become satellites because the latter is more
successfùl.
(b) Gonadal hvestment
Testes size differed significantly between parentals ( 1.82 *
0.56 g), satellites (0.62 * 0.19 g), and sneakers (0.20 0.10 g)
(ANOVA: Fzlw = 249.90, p < 0.001). The GSI of sneakers
(3.66 * 1.45) and satellites (3 -74 1 .O6) did not differ
(ANOVA: Ft = 106.44, p = 0.93), but both were significantly greater
than that of parentals ( 1.32 * 0.29) @ost hoc: p < 0.00 1).
There was a positive linear relationship between parental gonad
weight and body weight (r' = 0.60, d.f.
= 67, p < 0.00 1; Figure la). In cornparison, there was a
positive natural logarithmic relationship
between cuckolder male gonad weight and body weight (r2 = 0.83,
d.f = 73, p c 0.001; Figure
-
Ib). More specifically, there was a positive linear relationship
within sneakers (2 = 0.48, d.f. =
46, p < 0.001) and within satellites (2 = 0.54, d.f. = 25, p
c 0.001). The slope of the regressions,
or the rate of increase of gonad weight with body weight, was
significantly greater for sneakers
than satellites (t-test: t = 2.92, d.f. = 7 1, p c 0.005).
(cl Sperm morphologv
No significant difference was detected between the morphology of
parental and
cuckolder sperm (ANOVA - tail length: Fi,s = 0.37, p = 0.56;
head width: FiVs = 1.07, p = 0.33;
head length: FI,8 = 0.18, p = 0.68; mid-piece length: = 1.86, p
= 0.32; mid-piece width: Fi.s =
0.053, p = 0.83). There was little variation within the measured
morphological characten (rnean
coefficient of variance - cuckolders: 0.1 1 * 0.06; parentals:
0.16 * 0.07), and there was high repeatability among the three
measurements for each morphological chancter (the standard
deviations did not exceed 5 percent of the mean), which gives
confidence to the measurement
technique.
{d) Sperm Longevity
Parental sperm lived 40% longer than the sperm of cuckolders
(parentals: 159s k 30s,
sneakers: 109s * 19s, satellites: 120s * 33s; ANOVA: FZt3 =
7.66, p = 0.003; Figure 2a). There was no significant difierence
within cuckolden, between sneakers and satellites @ = 0.75).
(e) Spem quantiîy
In equal volumes of ejaculate, cuckolders had more spem than did
parentals. The
density of sperm produced by satellites exceeded that of
sneakers which in tum exceeded that of
-
parentals (satellite: 2 . 0 ~ 107 0 . 4 ~ 10' per mm3; sneaker 1
. 4 ~ 107 * 0 . 2 ~ 107; parental: 9 . 0 ~ 1 o6 k 1 .0~10~ ; ANOVA:
FZ,12d = 290.25, p < 0.001; Figure 26). Sperm density was
linearly related to
gonad weight in cuckolders (cuckolders: # = 0.57, d.f. = 46, p
< 0.00 1; Figure 36). Sneakers and
satellites had similar relationships (sneakers: 2 = 0.26, d.f. =
20, p = 0.02; satellites: r? = 0.50,
d.f. = 24,p < 0.001; t-tesr: t = 0.38, d.f. = 44, p >
0.50). There was no significant relationship in
parentals (7 = 0.0 L 5 , d.f. = 74, p = 0.29; Figure 30).
fl Ejaczrlate competitiveness
There was a significant positive correlation between patemity
and gonad weight in
cuckoldes (p = 0.82, n = 10, p = 0.004). There was no such
relationship in parentals (p = 0.26,
n = 10, p = 0.46 1). Since cuckolder gonad weight c m be used to
predict their ejaculate spem
density, the patemity of cuckolders is correlated with the
number of sperm present in the snipped
ejaculate.
-
4. DISCUSSION
We have tested 13 predictions from the literature and our
earlier work regarding the
mating opportunity and sperm investment strategies of the
alternative male mating tactics in
bluegill sunfish. Of these, 8 were confirmed and 5 were
refuted.
(a) Ciickolder mating opportunities
As predicted, small and large cuckolders had more mating
opportunity than did
intermediate sized cuckolden relative to their age-cohort size.
This may be because intermediate
size cuckolders could not sneak as well as smaller sneakers or
mirnic females as well as larger
satellites. In addition, we found that satellites were the most
successful individuals in gaining
access to matings and in obtaining reproductive success. This
confirms the assumption that
sneakers malce the transition to satellites because the latter
tactic fertilizes more eggs (Gross
1982). Disruptive selection that is associated with the extreme
body sizes of different mating
tactics is also found in salmon, where large males fight and
small males sneak (Gross 1985;
Flemming & Gross L 994). To our knowledge, this study is the
first to repon disruptive selection
within a single reproductive life history.
itals had absolu
(b) Gonadal investment
As predicted, paren itely larger gonads than cuckolders. This is
may be due
to the greater number of matings that parentals have access to
since they fertilize 4-7 times more
eggs than cuckolders. We can reject the notion that the
difference in testes size arnong
alternative tactics are due to allometry alone, since the
relationship between testes size and body
size is linearly discontinuous between sneakers, satellites, and
parentals. In addition, as
-
predicted, satellites have absolutely Iarger testes than
sneaken, possibly a result of the larger
number of matings that they have access to and the greater
number of eggs that they fertilize.
Gage et al. (1995) and studies in Tabonky (1998) have also shown
that guarding males have
absolutely larger testes than sneakers. The total number of eggs
available to males of alternative
tactics may explain the differences in absolute gonadal
investment between alternative mating
tactics.
As predicted by Parker's (1990, 1998) Sneak-Guard mode1 of
sperrn competition,
cuckolders have a larger GSI than parentals. For cuckolders,
Parker's prediction implies that the
return in reproductive success fiom investing in testes, and
therefore higher success in sperm
competition, outweighs the reduced retum fiom investing in soma
and rnating opportunity.
Although not confirmed in this study, Gross (1982) in fact found
that sneakers do have a greater
GSI than satellites, as predicted by their more disadvantaged
mating position (Gage et al. 1995;
Parker 1998). This suggests that the reproductive success of
sneaken Iargely depends on their
success under sperm competition in a fewer number of rnatings,
while that of satellites and
parentals rnay depend more upon their total mating
oppominities.
As predicted, parental testes size increased with body size
(also Gage et al. 1995,
Taborsky 1994). Contrary to prediction, testes size did not
decrease with body size in sneakers.
Since large sneakers have fewer mating opportunities than small
sneakers, we expected that the
testes of large sneaker would actually be smaller. The increase
in testes size with body size in
sneakers may be a result of adaptive compensation in gonadal
investment since meaken have a
disfavoured mating position under sperm competition (Parker
1990, 1998; Gage et al. 1995). In
addition, since large sneakers obtain few mating oppomuiities
than srnall sneakers, the value of
each mating increases as sneakers increase in body size. Thus,
Iarger sneakers may invest in
-
absolutely larger testes to increase their fertilization success
per mating. As predicted, large
satellites have larger testes than srnall satellites, but this
may be due to either their increased
mating opportunity. Overall, the relationship between testes
size and body size was logarithmic
in cuckolders, with satellites having a lower rate of increase
than sneakers. The difference in the
relationship between testes size and body size between the
alternative tactics could be a result of
the different renims in reproductive success gained by
alternative tactics in somatic versus
gonadal investment. Larger parentals invest in larger testes
since they obtain more eggs than
srnalier parentals (Gross & MacMillan 1984), and satellites
may have a lower rate of investment
in testes than sneakers since increased investment in somatic
development make them more
convincing female mimics.
(c) Sperm rnorpllology
Contrary to the prediction that cuckolders rnay have larger
sperm, there was no difference
in the sperm morphology of cuckolden and parentals. Although it
is thought that sperm size
should decrease with competition in extemally fertilizing fishes
(Stockley et al. 1997). a lack of
difference in the sperm size between alternative mating tactics
was also f o n d in Atlantic salmon
(Salmo salar, Gage et al. 1995) and bluehead wrasse (Thalassoma
bifmciatum, Schiirer &
Robertson 1999). This could indicate that the cross-species
trend in sperm size may not be
dependent on sperm competition but other correlated aspects of
mating systems such as egg size
or the ecology of the aquatic environment in which fertilization
takes place (e-g., Stockley et al.
1996; Petersen & Warner 19%).
(d) Sperm l o n g w
-
Contrary to the predictions that cuckolders have longer lived
sperm, as f o n d for the
precocious parr in Atlantic salmon (Gage et al. 1993, we found
that parental males have the
longer lived sperm. This contradiction rnay be due to
differences in the two mating systems. in
Atlantic salmon, both anadramous males and precocious parr rnay
experience similar intensities
and risks of sperm competition as several anadromous males and
precocious pan typically
participate in a single spawning (Fleming 1996). The precocious
parr in Atlantic salmon was
thought to have a disadvantaged mating position and this rnay
select for greater sperm longevity
(Gage et al. 1995). In contrast, parental bluegill experience a
relatively low intensity and risk of
sperm competition as compared to cuckolders. The longer
fertilization window available to
parentals rnay select for the release of relatively fewer longer
lived sperm. Cuckolden, since
they always experience sperm competition, rnay release more
shorter-lived sperm. As there rnay
also be a trade-off in sperm longevity and speed, cuckolders
rnay produce faster sperm that could
also out-compete parental sperm by fertilizing eggs faster
(Levitan 2000). Sperm speed,
however, remain to be investigated. Overall, we propose that
longer lived spem is not
exclusively a characteristic of sneaker tactics, but that sperm
longevity may be correlated with
the availability of unfertilized eggs.
(el Sperm quantity
As predicted, cuckolders have greater ejaculate spem density
than parentals. This is
consistent with the finding in bluehead wcasse (Schiirer &
Robertson 1999), where the
cuckolder-like initial-phase (P) males have greater ejaculate
sperm density than the parental-like
terminal-phase (TP) males. The higher spem density of cuckolders
may be due to their
limitation in body cavity for sperrn storage (Gross 1982) or
selection for the ejaculation of sperm
-
at a greater rate when under spenn competition (suggested by
Schiirer & Robertson 1999).
Contrary to prediction, satellites had greater ejaculate sperm
density than sneakers. This may be
because, given their decreased investrnent in gonadal
development, satellites require a tighter
packaging of sperm so as to meet the demands on sperm output for
the four times more mating
opportunities that they obtain.
(B Ejacdute competitiveness
As predicted, we found that fertilization success was correlated
with the number of sperm
released in an ejaculate. Although this result is only
explicitly demonstmted in cuckolders, the
same should hold tme for parentals, although we were not able to
manipulate their ejaculate
spenn number. Since the fertilization experiment eliminates the
effect of proximity, it
demonstrates that ejaculate sperm number plays a role in
determining a male's fertilization
success under spem competition. This confirms that sperm
competition in extemally fertilizing
species may operate as a raffle, which has been assumed in
several theoretical models (reviewed
in Parker 1998).
-
ACKNOWLEDGEMENTS
We thank Silvia D'Arnelio, Luca Cargenelli, Tamara Janasik, Anna
Lawson, Tracy Michalak,
and Jemifer Moran for help with data collection, and Bruce Smith
of Ithaca College for the use
of his microscopy equipment. This work is supported by gants
from NSERC of Canada.
-
REFERENCES
Bouola, J. J. & Russell, L. D. 1992 Electron microscopy:
principles and techniques for
biologists. Boston, Massachusetts: Jones and Bartlett
Publishen.
Chakraborty, R., Meagher, T.R., Smouse, P.E. 1988 Parentage
analysis with genetic markers in
natural populations. 1. The expected proportion of offspring
with unambiguous paternity.
Genetics 118,527-536.
Colbourne, J. K., Neff. B. D., Wright, J. M., & Gross, M. R.
1996 DNA fingerprinting of bluegill
sunfish (Lepomis macrochirus) using (GT), microsatellites and
its potential for assessrnent of
mating success. Con. J. Fish. Aquat. Sei. 53,342-349.
Flemming, I.A. 1996 Reproductive strategies of Atlantic salmon:
ecology and evolution. Rev.
Fish. Bio. Fish. 6 , 379-4 16.
Flemming, I.A. & Gross, M.R. 1994 Breeding cornpetition in a
Pacific salmon (coho:
Oncorlynclizis kisutch): measures of n a m l and sexual
selection. Evoliction 48,637-657.
Gage, M . J. G., Stockley. P., & Parker, G. A. 1995 Effects
of alternative mating strategies on
characteristics of spem production in the Atlantic salmon (Salmo
salar): theoretical and
empirical investigations. Phil. Tram. R. Soc. Lond. B
350,391-399.
Gage, M. J. G., Stockley, P., & Parker, G. A. 1998 Sperm
morphometry in the Atlantic salmon.
J. Fish Biol. 53,835-840.
Gross, M.R. 1980 Sexual selection and the evolution of
reproductive strategies in sunfishes
(Lepomis: Centrarchidae). PhD. thesis, University of Utah,
Utah.
Gross, M. R 1982 Sneakers, satellites, and parentals:
polymorphic mating strategies in North
Amencan sudishes. 2. Tierpsychol. 60, 1-26.
Gross, M.R 1985 Disruptive selection for alternative life
histones in salmon. Nature 313,47-48.
-
Gross, M. R. 1996 Alternative reproductive strategies and
tactics: diversity within sexes. Trend.
Ecol. Evol. 11,92-98.
Gross, M. R. & Chamov, E. L. 1980 Alternative male life
histores in bluegill sunfish. Proc. Nad
Acad. Sci. USA 77,6937-6940.
Gross, M. R. & MacMillan A. M. 198 1 Predation and the
evolution of colonial nesting in
bluegill sunfish (Lepomis macrochinu). Behav. Ecol. Sociobiol.
8, 1 63- 1 74.
Hyat, M. A. 1974 Principles and techniques of scanning electron
microscopy. Vol. 2. New York:
Van Nostrand Reinhold Co.
LaMunyon, C.W. & Ward, S. Evolution of sperm size in
nematodes: sperm competition favours
larger sperm. P