Content uploaded by Joel C. Trexler
Author content
All content in this area was uploaded by Joel C. Trexler on Oct 21, 2014
Content may be subject to copyright.
Nordic Society Oikos
Size-Selective Predation of Sailfin Mollies by Two Species of Heron
Author(s): Joel C. Trexler, Robert C. Tempe and Joseph Travis
Source:
Oikos,
Vol. 69, Fasc. 2 (Mar., 1994), pp. 250-258
Published by: Wiley on behalf of Nordic Society Oikos
Stable URL: http://www.jstor.org/stable/3546145 .
Accessed: 03/06/2013 11:41
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
Wiley and Nordic Society Oikos are collaborating with JSTOR to digitize, preserve and extend access to Oikos.
http://www.jstor.org
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
OIKOS 69: 250-258.
Copenhagen
1994
Size-selective
predation
of sailfin
moshes
by
two
species
of
heron
Joel
C. Trexler,
Robert C. Tempe
and
Joseph
Travis
Trexler,
J. C., Tempe,
R. C. and Travis,
J. 1994.
Size-selective
predation
of sailfin
mollies by
two
species
of
heron.
- Oikos
69: 250-258.
Two species
of
herons
(Aves:
Ardeidae),
the great
egret
and the snowy
egret,
fed
preferentially
on relatively
large
individual
sailfin
mollies (Pisces:
Poeciliidae).
The
level of
selectivity
by great
egrets decreased
when
plant
cover was
added to
the
experimental
conditions.
We
documented
the
preference
of
snowy
egrets for
large
mollies
in all combinations of
two
prey
sex ratios and
two prey
size distributions
with
50%
plant
cover.
Snowy egrets
chose
females
preferentially
to males
when
females
were rare,
but not
when they
were
commmon. This
preference
may be
caused by
the
larger overall
size of females
when gravid
in
comparison
to
males of
the same length
and/or by
the behavior
of
females
when they
are
in
a minority relative
to males.
Selective
predation
by wading
birds
is
likely
to contribute
to
patterns
of size
variation
among
some
populations
of sailfin
mollies but
not
to
the
patterns
of
female-skewed
sex
ratios in
adults.
J. C. Trexler, Dept
of Biological
Sciences,
Florida International
Univ.,
University
Park,
Miami,
FL
33199,
USA.
- R.
C.
Tempe,
Inst.
of
Marine
Science,
Univ.
of
North
Carolina,
3407 Arendell
St.,
Morehead
City,
NC
28557,
USA.
- J.
Travis,
Dept of
Biological
Science B-142,
Florida
State
Univ.,
Tallahassee,
FL 32306-2043,
USA.
Size-selective
predation
by wading
birds
can
affect
the
sex
ratio,
size-structure,
and
local
distribution
of
fish
populations
(Britton
and
Moser 1982,
Power et
al.
1989).
The sailfin
molly,
Poecilia latipinna
(Poeciliidae),
a
small
euryhaline
fish,
exhibits
variation
among
local
pop-
ulations
in
the size
distribution
of both
males
and
females
(Snelson
1985,
Farr
et al.
1986,
Trexler
1986,
Travis
and
Trexler
1987).
It also exhibits
female-biased
adult
sex
ratios
despite
an
even
secondary
sex
ratio
(Snelson
and
Wetherington
1980).
In
this
paper
we
report
several
ex-
periments
designed
to examine
the validity
of the
hy-
potheses
that size-
and
gender-specific
predation
by
wad-
ing
birds
could
promote
differences
among
populations
in
body
size distributions
and
female-biased
sex
ratios
among
adults.
Differences
among
molly populations
in
size-structure
are
large;
average
body
sizes of
mature
males
can
range
among
populations
from
21 to
38 mm
standard
length
(defined
below)
and
the
shapes
of male
body
size
distri-
butions
can
range
from
uniform
to normal
to
nearly
lognormal
(Farr
et al. 1986,
Trexler
1986, Travis
and
Trexler
1987,
Travis
1989).
Small
male mollies
(< 30
mm
standard
length)
occur
in
nearly
all
populations,
and
the
major
contribution
to
the
differences
in size
distributions
among populations
is
the
presence
or
absence
of
larger
males.
Average
size
of
females
varies
in
parallel
to that
of
males,
even
with
respect
to the
ubiquity
of small
females
and
the "patchy"
distribution
of larger
ones (Travis
and
Trexler
1987).
Phenotypic
plasticity,
direct
environmen-
tal effects
on
body
size,
cannot
account
for
the
levels
of
observed
variation
(Trexler
and
Travis
1990,
Trexler
et
al.
1990).
Variation
among
populations
in male
size
reflect
genetic
variation
in size at maturity,
because
males
do
not
grow
appreciably
after
maturity (Snelson
1984,
Travis
et
al. 1989)
and
variation
in size
at maturity
is highly
heri-
table
(Travis
et al.
1993).
Variation
among populations
in
female
size may
reflect only
small
differences
in size
at
maturity
(Trexler
and Travis 1990,
Trexler
et al. 1990)
but
larger
differences
in age
structure
due
to
postmaturation
growth
and survival
rates
(Trexler
and Travis
unpubl.).
Accepted
19 July
1993
Copyright
?D
OIKOS 1994
ISSN
0030-1299
Printed
in Denmark
- all rights
reserved
250 OIKOS 69:2
(1994)
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
There are
several
possible
explanations
for these
patterns
invoking
natural selection
(Trexler
et al.
1992),
including
size-specific
predation
by
wading
birds.
Sailfin
mollies
are
common prey items of several
species
of
wading birds
(Ogden et
al. 1976,
Kushlan
1979,
J. Kushlan and
P.
Kushlan,
pers.
comm.)
and
wading
bird
predation
is usu-
ally
size-specific (above
references).
Wading
bird
preda-
tion
could
account for not
only
the differences in
average
size
but
also the
major
reason for those
differences,
the
presence
or absence of
larger
fish.
Wading
bird predation
is a strong
candidate for
an
agent
of differential
mortality
between
genders
that
would produce female-biased sex ratios. Male mollies
spend a great deal
of
time
actively
pursuing
and
courting
females, and male
display
rates increase
allometrically
with
increases
in
male standard
length (Baird
1968,
Farr
et al. 1986,
Farr and Travis
1986,
Travis
unpubl.).
Male
secondary
sex
characters such as dorsal
fin
size
also
increase
allometrically with increases in
standard
length
(Snelson
1985,
Farr et al. 1986).
The
combination of
phenotypic features and
activity
levels of
larger
males
may make them
more
susceptible
to
the visual
foraging
of
wading birds
(Kushlan
1978).
In other
species of
poeci-
hid
fishes,
variation
among locations
in
the
intensity
of
predation
by
visually
oriented
predators
has been shown
to be
associated
with
male color
patterns
(Endler
1980),
body
size
(Reznick
1982,
Reznick
and Endler
1982),
rates
of
courtship
displays
(Farr
1975,
Breden
and Stoner
1987,
Endler
1987, Stoner
and Breden
1988), and school-
ing
behavior
(Liley
and
Seghers
1975, Breden et al.
1987,
Magurran
and Seghers
1990).
The
influence of
wading
bird
predation
on
size-struc-
ture and
sex
ratio
may
not
be independent
in
molly
populations.
Sailfin
mollies are
one of the
very
few
spe-
cies of
poeciliid fishes in
which
males can
equal or
exceed the sizes
of females
because of the
wide
range of
sizes
at which
males
may mature
(Farr
1989).
Preferential
predation on
larger
mollies
may
actually
result from
pref-
erential
predation of males
because
of
gender-specific
attributes such as
color
or
behavior.
Experiments
must be
designed to
separate
the
potential
confounding of
individ-
ual
body size
with
gender. In
this
paper
we
report the
results of
experimental trials
of
wading bird
predation
on
sailfin
mollies
designed to
evaluate
whether
predation is
size-
and
gender-specific
and
whether the
relative
risk of
predation is a function of
the
overall
shape of the
size
distribution of
prey
available or
the sex
ratio of
prey
available.
Materials
and
methods
Experimental
conditions
We
studied
feeding
preferences in
two
species of
wading
birds
during
the
spring
of
1987,
summer
1989, and
spring
1990.
In
1987,
a
great
egret(s)
(Casmerodius
alba)
served
as
the
predator, while
snowy egrets
(Egretta thula)
were
studied in 1989 and
1990. The
experiments
were
con-
ducted at the
margin
of
a
brackish
water
pond
located
on
the
northern
edge
of
Tampa
Bay,
Pinellas
County,
Flor-
ida,
USA. The birds
studied were
free-ranging, they
came
and left the
study
area as
they
wished.
To
our
knowledge,
they
were not
habituated to
feeding by
humans,
and
they
flew
away when we
approached.
Egrets
establish
feeding
territories to which
they
return and from which
they
chase
away
other birds
(Hancock
and Kushlan
1984).
We
did
not
attempt to mark the birds we
studied,
we
did not wish
to
disturb them
by
capture
and
handling,
so we do
not
know
the
identity
of
the
bird(s)
in
our
feeding
trials.
Thus,
it
is
possible
that
only
one individual bird
participa-
ted
in
our
experiments
in
each of the
first two
years,
1987
and 1989. We know that at least two different
snowy
egrets
participated
in
the 1990
experiment
because
two
were
seen
simultaneously
in
our
study
area. Trials where
more
than
one bird entered the
pool
to
feed were elim-
inated
from
analysis.
We
decided not to use
caged
birds to avoid
artifacts
possible when birds are fed
on a regular
schedule.
Our
birds fed in
areas
they
chose
because
of
natural fish
availability
at times
they
chose.
To
gain
this
perspective,
we were forced to
sacrifice
knowledge
of the
identity
of
the birds we studied
and the
number
of
birds studied.
All experiments were
conducted
by
placing
sailfin
mollies
of
known size
in
a
wading pool
(1.5
m
diameter).
The
pool
was
placed
either
in
the water
or on the
edge
of
a
large
pond
where
several
species
of
wading
birds
com-
monly
fed.
Water from the
pond
was
used to
fill
the
wading pool to a depth
of
approximately
10 cm.
The
bottom of
the
pool
was
painted dark
gray
with
a
non-toxic
plastic
paint to match the
natural
color
of
the
pond
bottom
as closely
as possible.
In 1987, experiments
were run
either
without
floating plant
cover for the
fish or
with
50%
cover
of
floating
artificial
aquarium
plants.
Fifty-
percent
cover
was
provided
in
all
cases
in
the
1989 and
1990
experiments.
These
experimental
conditions
are
within
the
range of
conditions
where
mollies
are
found
naturally. For
example, in
Florida
mollies can
be
found in
barren
pools
above
neap
high tide
and,
during
the dry
season,
in
solution
holes
containing
clear water
and no
hiding
places
where
they
may
be
trapped until
the
pools
completely
dry
or
are
refilled by
spring
rains.
The
experi-
ments
were
designed to
provide
insight
into
the
effect of
different
levels of
habitat
complexity
or, in
the
1989-90
experiments,
to
hold
complexity
constant.
Of
course,
mollies can
also
be found
in
highly
complex
habitats
with
many
hiding
places
such
as
those with
100%
plant
cover.
We
would
not
expect
bird
predation
to be
an
important
source of
mortality
in
such
places
and the
logistics of
obtaining
proper
combinations
of
size and
sex of
fish
precluded
our
examination
of such
conditions.
Birds
which
feed
by
touch, such
as wood
storks
(Mycteria
americana),
are
potentially
size-selective
predators of
fish even
when no
visibility is
possible. This
remains to
be
explored.
All
fish
were
measured
at the
outset
of the
experiment
OIKOS 69:2 (1994) 251
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
Table
1.
The relative
abundance
of each
size and sex of fish
in
experiments
conducted
in 1989
and
1990.
The
actual
number
of
fish
in
each
category
at the
outset
of the
experiment
is
listed
in
parentheses.
Size Sex Total
Rare sex Common
sex
Uniform size distribution
Small 0.083 (2) 0.250
(6) 0.333
Medium 0.083 (2) 0.250
(6) 0.333
Large 0.083 (2) 0.250
(6) 0.333
Total 0.250 0.750
Skewed size
distribution
Small 0.125 (3) 0.375
(9) 0.500
Medium 0.083 (2) 0.250
(6) 0.333
Large 0.042 (1) 0.125
(3) 0.167
Total 0.250 0.750
and placed
into
one
of
three
size groups
based
on stan-
dard length (the
length of
the dorsal
surface
from
the tip
of the
mouth to the
point
of insertion
of the caudal
fin
rays on
the
caudal
peduncle):
small
19.5
- 30 mm, me-
dium 30.5 - 40 mm, large
40.5 - 65.5 mm. It
is impos-
sible to
distinguish
sexually
immature large
males from
non-gravid
and
non-sexually
receptive
females,
so some
immature
males
may
have been
classified
as females.
If
there are differences
in
preference
between
females and
juvenile
males of the
same
size,
this pooling
renders our
ability
to detect preferences
for females
conservative.
Twenty
fish
were
placed
in
the wading pool
at the outset
of
each feeding
trial
in 1987 and
24 fish
were used
in
1989
and 1990
studies.
Birds
were
permitted
to feed
until
approximately
one-third
of
the fish
in the
pool
were eaten
based
on
direct observation
of
the number
of fish eaten
in
a
feeding
bout.
When this
point
was
reached,
the bird
was
chased
from the
pool and
the
remaining
fish were
mea-
sured
and
their sex
was
noted.
No more than
three feed-
ing trials
were
conducted
each day
to
minimize the
poten-
tial
of
predator
satiation.
Experimental
design
In 1987 male and female
fish
were
placed in pools in
frequencies
dictated
by
their
availability.
Data
from these
trials
permit
us to test
for size-selective
feeding
but
not to
separate
any
such
selectivity
from
gender-dependent
prey
selection
because
the
sample
sizes of
each sex
were
not
controlled.
Eight
feeding
trials
were conducted
without
plant
cover
present
and
eight
trials
were conducted
with
plant
cover
present.
In 1989 and 1990 we designed
a substantially
more
complex
experimental
protocol
intended
to
separate
se-
lection
based on size from
that
based on gender.
We
tested these
preferences
under
a uniform
size distribution
and a skewed
size distribution
(more small
fish than
large,
as is often
observed
in natural populations)
and
male
and female-biased
sex ratios
(75% male and
25%
male) (Table 1). Four trials
were
conducted
for each
of
the four
combinations
of fish size and sex
ratio in
each
year
for a total
of eight
replicate
trials of
each type
or
twenty-four total
trials.
Our experimental
design is
based
on
the null hypothesis
that
if no
feeding
preferences
are
present,
the different
sized
fish and
the two
genders
will
be eaten at the
frequency
at which
they occur
in the
pool.
If the
frequency
of different
sized individuals or
members
of the two sexes in the
bird's
diet deviates
from
their
relative
abundance,
we reject the
null hypothesis
and
accept
the alternative
that
the birds
are demonstrating
a
feeding
preference.
The
source of
bird preferences
are
not
directly tested
(e.g., fish
behavior,
color
patterns,
visual
acuity
of
the birds)
by
any one
of the four
experimental
conditions
and
is not the
focus of
this
research;
we
only
seek to determine
if size or sex-specific
mortality
are
potentially
induced by
wading birds.
However,
the
pat-
tem of results across the
four experimental
designs
may
provide
insight
into such issues.
Data analysis
The analysis
of prey-choice
experiments
is not straight-
forward
for two reasons.
First,
consumption
rates
of dif-
ferent
prey types
in a trial are not
independent
(Hay
et
al.
1988,
Peterson
and
Renaud 1989).
Second,
when the
total
number
of prey
is not
large relative
to the predator's
consumption
rate,
the predator
is sampling
without
re-
placement.
This
means that the
probabilities
of
removal
for each
prey
category
may
change during
the
individual
trial.
Multivariate analysis
of variance methods
(Roa
1992) address
the first problem
but not the
second.
We used
the
Chesson-Manly
estimator
of
selectivity
to
document
the
presence
or absence
of
prey
preferences
(Chesson
1983, Manly
1985).
This
estimator
is calculated
for each prey
category
in each trial,
and can be in-
terpreted
as the
probability
that
a prey
item in
that
cate-
gory
will
be selected
first
by
a predator
if
there
were
an
equal number
of items
in all categories.
In
other
words,
the index
estimates
the
probability
that
an item in
cate-
gory
i will be chosen
first if the encounter
rates
for all
categories
were
equal.
The behavior
of the
index of selectivity
is straight-
forward
(although
its theoretical
derivation is
not).
If no
preferences
exist,
then
the estimator
for
each category
will
be
equal
to l/n
where n
is the number
of
categories.
Note
that
this is
true
even
when
the
encounter
rates
in
the
experimental
trial
are not initially equal among
cate-
gories.
This
property
allows
the
estimator
to be used
in
experimental
designs
that
purposefully
vary
encounter
rates.
If category
j is preferred,
then
its value for the
selectivity
index
will exceed l/n.
When
only
two
cate-
gories
exist
the value
for the
non-preferred
category
will
be 1
minus
the
value
for the
preferred
category,
because
the
estimators for each
category
are constrained
to sum
to
252 OIKOS 69:2 ('1994)
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
Table
2. The
median
values
of
the Manly-Chesson
index
for
each combination
of
size
and
sex
of fish
in
1987
experiment.
Size
class Without
cover With
cover
Small 0.033 0.243
Medium 0.191 0.254
Large 0.634 0.510
1 (recall
the
interpretation
as a probability
of
selection
given
equal
encounter
rates).
In general,
preferred
cate-
gories
have
higher
values
for the index
and
unpopular
categories
have
lower values
with
the constraint
that
the
sum
of all
values
must
be 1.
When
sampling
is
without
replacement
the
appropriate
estimator
is
ln((nj,,-rj)1nj,)
Oti
= ,~ =1 .,m(1
E ln((nj,-rj)Inj,0)
j=1
where
ri
is the
number
of food
items
of
type
i
eaten,
nio
is
the
number
of
food items
of
type
i
available
at the
outset
of
each feeding
trial,
and
m is
the
number
of
food
types
present
(Chesson
1983).
In
about
25%
of
the
cases,
the
numerator
of
eq. (1)
was replaced
with
the value
ln(O.05)
because
all
prey
in a category
were
consumed
and
this
value
was also used to
obtain
the
summation
in
the
denominator.
The
use of
nonparametric
hypothesis
tests
(see
below)
eliminates
potential
artifacts
in the
value
of
the
estimates
that could
be
produced
by
this substitution.
We used
these
estimators
to
test
the
null hypothesis
of
no preference
against
several
a priori
alternatives
with
Wilcoxon's
signed
rank test
(Hollander
and
Wolfe
1973).
In each case
we
tested hypotheses
based
on n-
I
categories
of
prey
type,
where n is
the number
of
categories
avail-
able.
This
avoids
problems
of
independence
in
statistical
tests that
result from
n-
I of
the
categories
determining
the
results of
the nth category
(Chesson
1983).
When testing
for selectivity
based
on
the
sex of
the prey,
we used
Wilcoxon's test
for
one sample
data
(Hollander
and
Wolfe
1973:
50-53)
to
test
the
null
hypothesis
that
males
and
females
were
equally
likely
to
be
preyed
upon.
In
other words,
we tested
the hypothesis
that
cai
for males
did
not differ
from
0.5. When testing
for
selectivity
based
on
size
we used Wilcoxon's
test
for paired
replicates
(Hollander
and
Wolfe
1973:
27-28)
to
test
that
(xi
for
small fish
was
equal
to
ai
for
large
ones.
Given
that
three
categories
of size
were used
in these
experiments,
the
expected
value
of
(xi
for
each
size
category
was
0.33.
Variation
in the
number
of
medium-sized
fishes
eaten
will affect
the absolute
value
of
alpha
obtained
from
feeding
trials but
has no effect
on
the ratio
of
alphas
calculated
for a
given
experimental
trial.
Use
of nonpara-
metric statistics
eliminates
any
artifactual
results
caused
by
changes
in the absolute
values
of
the
alphas
because
test
statistics
are
based
on
the
number
of
trials
where
the
rank
of the
alphas
for
small
and large
fish
are
the
same.
Our
analysis
protocol
for
the
1988-1989
experiments
does
not
test
for
all
possible
types
of
feeding
selectivity.
Instead,
it tests
the
a priori
hypothesis
of directional
selection
favoring
large
or
small
fish;
the
possibilities
of
disruptive
or
optimizing
selection
are
not
tested.
There
is
a statistical
and
a biological
reason
for
our
restriction
of
alternative
hypotheses.
The statistical
reason
is that
by
Without
Cover
1.0
0.5
a.
0.0
-0.5 X
small medium large
SIZE
50% Cover
1.0
0.5
a.
oc
0.0
-0.5 |
small medium large
SIZE
Fig.
1. Selectivity
index
(a()
for
each
size
class
from
1987
study.
The
points
connected
by
each
line
are
the
results
for
one
experi-
mental
trial
and
are
not
independent.
OIKOS 69:2 (1994) 253
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
Table 3. The median values of the Manly-Chesson
index for
each combination of size and sex of fish in 1989
and 1990
experiments. Rows and columns do not sum to totals.
Size Sex Total
Male Female
25% males with uniform size distribution
Small 0.000 0.004 0.014
Medium 0.047 0.020 0.104
Large 0.394 0.394 0.796
Total 0.584 0.416
25% males with skewed size distribution
Small 0.025 0.008 0.035
Medium 0.043 0.027 0.068
Large 0.446 0.446 0.892
Total 0.496 0.504
75% males with uniform size distribution
Small 0.030 0.033 0.065
Medium 0.087 0.071 0.170
Large 0.140 0.654 0.748
Total 0.231 0.769
75%
males with skewed size distribution
Small 0.025 0.037 0.061
Medium 0.066 0.316 0.365
Large 0.160 0.349 0.593
Total 0.276 0.724
testing only
these a priori
hypotheses,
we increase
the
statistical
power
of
our
approach.
The fixed number of
fish
per
size and sex
category
limits our
options
for tests
of these
patterns
of selection to
multiple
tests
comparing
the small vs medium and
large
vs
medium and
adjusting
for
the family error rate. Biologically,
there is no
reason
to expect anything
other
than
directional
selection
by
wading
birds on sailfin
molly
sizes (see references
to
wading
bird
foraging patterns
in Introduction and Dis-
cussion
sections). Moreover, disruptive
or
optimizing
se-
lection could
not
explain
directional
differences
in
aver-
age
fish size
among populations,
which is the
pattern
we
seek to understand.
Visual examination
of the results
support
the
decision
to
test
only
for
directional
selection;
the
median
alphas
for
the medium size
class
are
interme-
diate between
those
of
small and
large
in
all cases
(Table
3) and
plots
of
the
results
from
individual
experiments
provide
no indication of
optimizing
or
disruptive
selec-
tion
(Figs 1 and
2).
Results
Great
egret feeding
trials
The great egret(s)
demonstrated
a preference
for
large
fish
when there
was
no
place
for
the fish to hide
(i.e.,
no
floating plants
in
a bare
pool) and less
preference
when
hiding places for fish were
provided (Table 2, Fig. 1).
Large
individuals
were
eaten
at a greater
frequency
than
small ones in all eight feeding trials without cover. The
null hypothesis of no
preference
for small versus
large
fish was rejected (T+8
= 36, P = 0.008). The feeding
preference for large fish was not demonstrated when
50%
cover was added to the
pool,
although the data were
suggestive of such preferences. We
failed
to reject the
null hypothesis of no preference for small versus large
fish
(T+7
= 23, P
=
0.078). Large individuals were chosen
preferentially over small ones in six of the eight trials
with plant cover, but this was not adequate to reject the
null hypothesis. The frequencies of large and small fish
eaten were equal in one of the remaining two trials so it
was excluded from the hypothesis test (Hollander and
Wolfe
1973: 28).
In
the other trial the frequency of small
fish in the bird's diet exceeded that of large fish. In this
trial,
oti
was
0.73 for small fish but only 0.16 for large
ones. Thus, cover decreased the effective selectivity of
foraging by the great egret(s) though it appears that these
birds prefer larger mollies when they can catch them.
Snowy egret feeding trials
The snowy egrets showed a preference for larger fish
without regard to gender under all experimental condi-
tions
(Ho:
Pr(large fish selected)
= Pr(small fish selected)
- 25% male with uniform size, T+8 = 36, P
=
0.008; 25%
male with skewed size, T+7
= 28, P = 0.016; 75% male
with
uniform size, T+8
= 35,
P
= 0.016;
75%
males
with
skewed size, T+8
= 36,
P = 0.008).
This
preference was
quite marked, particularly
so
for
the trials
involving
the
skewed size distribution and female biased
sex ratio in
which there were very few large
fish
to find (Table 3, Fig.
2).
In
contrast, a preference
for females
occurred only
in
the trials
in which females were rare
(Ho:
Pr(male chosen)
= 0.5
-
75% males
with
uniform and
with
skewed
size,
T+8
= 36,
P = 0.008). When
females were
the
common
gender,
birds
took
males
and
females
indiscriminately
(HO:
Pr(male chosen)
= 0.5 -
25% males
with uniform
size, T+8
= 21,
P = 0.371;
25% males
with
skewed
size,
T+7 = 10,
P
> 0.50; Fig.
2: note the different scales
on
the
Y-axis for
graphs
of
the
male
and
female
results).
The
preference
for
females
when
they
were rare was
strong
and unaffected
by
the
shape
of
the
size distribution.
There
is
no
reason
to
expect
that a
preference
for females
when
females were common would
emerge
from an
increased
sample size;
the
selectivity
values for
each
gender
were
very similar,
so the lack
of
significance
is not
likely
to be
a problem
of statistical
power.
It is possible
that
the
preference
for
larger
fish
is less
strong
when females
were rare
and the size distribution
was skewed
(Table 3).
The
selectivity
index
was
noticeably
lower
for
large
fish
in
this
treatment combination
and
it is
possible
that
an
increased
sample
size
would
reveal
such an
effect. The
conclusion
we draw from
these trials
is that size-selective
predation
occurred without
regard
to size-structure
or sex
ratio,
but
gender-specific predation
occurred
only
when
254 OIKOS 69:2 (1994)
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
75% Males and Uniform
Size Distribution
0.3 . 0.8
MALES 0.6 FEMALES
0.2
0.4
9L0.1
* .
0.2
0.0 0.0
-0.1 -0.2
small medium large small medium Urge
SIZE SIZE
75%d Males and Skewed Size Distribution
0.4 *0.6
MALES 0.6 ~~~FEMALES
0.3 MAE06
0.4
0.2 - 25
0.2
0.1 0.0
0.0 -0.2
small medium large email medium large
SIZE SIZE
Fig.
2.
Selectivity
index
((xi)
for each size
class
reported
separately
by
sex
for results
from
1989-1990
study.
The
points
connected
by
each
line are the results
for one
experimental
trial and are not
independent;
note that
each
line
on
panels
on the
left
(male
data)
has
a
counterpart
on the
right
panel
(female
data)
that was obtained
from the
same
trial.
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
25x. Males and Uniform
Size Distribution
0.8 0.8
0.6 MALES 0.6 FEMALES
0.4 0.4
-I~~~~~~~~~~~~~~~~~~~~~~~I
0.2 0.2
0.0 0.0 W
-0.2 -0.2
small medium large small medium large
SIZE SIZE
25% Males and Skewed Size Distribution
0.5 *0.5III
0.4 MALES 0.4 FEM
LES
0.3 0.3
Ai 0.2 IL 0.2 -
0.1 0.1 6
0.0 0.0
-0.1 -0.1L
small medium large email medium large
SIZE SIZE
Fig.
2,
continued.
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
females were
rare,
and
was
unaffected
by
the
shape
of
the
size distribution.
Discussion
We found consistent selection of
larger
prey by
wading
birds across two distinct distributions
of
body
size and
two
sex-ratios. This
consistency suggests
that
size-selec-
tive
foraging
on mollies
by
wading
birds
in
nature
will
be
robust
with
respect
to wide variation
in
these
parameters.
However, this does not
imply
that
wading
birds
will
play
effective roles
in
determining
the
size-structure of
molly
populations
in all
locations.
These birds
rely
on vision to
find
prey (Kushlan
1978)
and
decreased water
clarity
and
increased
vegetation
cover decrease foraging
success
(Custer and Osborn
1978,
this
study).
In addition,
their
effectiveness decreases as water
depth
increases
(Kushlan
1976, Custer and Osborn 1978, Kramer et al. 1983).
Thus, only
in
particular
types
of
locations
will
predation
by
wading
birds
be the
key
selective factor that deter-
mines the size-structure of
molly populations.
The Shark River
Slough
drainage
of the
Everglades
in
Florida
provides
an
example
of
the
type
of
location where
wading
bird
predation
may
be an important
selective
factor
for
molly
size
distributions.
Although
the
species
is
quite abundant in this
location, only
small
mollies can
commonly
be found there
(Loftus
and Kushlan 1987,
Trexler
pers.
obs.).
Numerically,
mollies
comprise
about
30% of
the diet of
great
egrets
(C. alba), 17% of the
diet
of
snowy egrets
(E. thula),
and 20% of
the diet of
tricolor
herons
(Egretta
tricolor)
in
nearby
habitats (J. Kushlan
and P. Kushlan pers. comm.). During the seasonal
droughts
in Shark
Slough,
fish
are
crowded into
small,
shallow
pools
where
large
numbers of
wading
birds
con-
gregate
and feed.
Large mollies are
common
in
nearby
habitats and there is no physical
barrier
to their
colo-
nization of
the
Slough
area
during the
wet season.
Thus,
wading
bird
predation
during the
dry
season may
be the
key
selective factor
that
maintains the
limited size
distri-
bution
seen
in Shark
Slough.
Additional
evidence sup-
porting
this
hypothesis
is
that
large mollies
can be
found
in
Shark
Slough
during the
dry season,
but only in
iso-
lated permanent
deep water habitats
such as large
solution
holes and
borrow
pits
where
wading
birds are
probably
less effective
predators (Trexler
pers.
obs.).
In more
general
terms,
wading bird
predation
is only
one of several
selective
agents
that
act to determine
variation in
the
size-structure of
sailfin
molly
populations
throughout
their
range. Mollies thrive in locations that
vary
widely
in
water
depth,
water
clarity, extent
of vege-
tation
cover,
and many
other
variables.
The variation
among populations
of mollies
in size structure
in the
eastern
half
of
their
range
is not
associated with
variation
in
either
any
one of
these
variables or
any
multivariate
combination of
them
(Trexler
1986,
Travis
and
Trexler
1987). Different
combinations
of causal factors
act in
different locations and the accumulated data indicate
that
there is no single explanation
for variation
in
the
size-
structure of these
populations.
The
results
reported
here,
in conjunction
with what is known about
wading
bird
foraging
success,
show that
wading
bird
predation
can be
the critical factor
in
specific types
of locations.
Contrary
to
expectation,
wading
bird
predation
is un-
likely to
produce
the
female-biased sex ratios
commonly
seen in
molly populations (Simanek
1978,
Snelson
and
Wetherington
1980).
The selective
predation
on
females
we observed
may
be a reflection of
simple size-specific
foraging
because
gravid
females
appear
wider
(and
thus
larger)
when
viewed from above
(as by
a
bird)
than males
of
the
same
length. Many
females used
in
our
study
were
gravid.
We
measured five females before and
after partu-
rition and found that
they
decreased in width
by 20%
after birth.
Thus,
our measure of female size based on
standard
length may
have under-estimated size as per-
ceived by
foraging
birds.
However, this does not explain
why the preference for
females
only appeared when
they
were the rare sex. We have
observed that when females
are
rare,
especially large
females,
the attention of
male
mollies focusses on them. In such cases, the females
swim
vigorously away
from the
groups
of
chasing
males,
possibly
drawing
the
attention
of
feeding
birds. We ex-
pected
that males would
be preferred
because their con-
spicuous
secondary
sex
characters and behavior
patterns
would
make them more apparent
to the birds, yet in
no
case were
males eaten at rates
above
their
relative abun-
dance.
Acknowledgements
- We thank J.
Reynolds and Eckerd
College,
St.
Petersburg, Florida, for
providing logistical and moral
sup-
port
for this
research. M.
Meyer
conducted the
field studies for
the
1987
experiment. We thank J. and P.
Kushlan for
sharing
information
on heron feeding and K.
Spitze and
P.
Stoddard
for
commenting on a penultimate
version of the paper. Our
work
was
supported financially by NSF grant BSR 88-18001
and
RCT was
partially supported by the Ford
Foundation program
at
Eckerd
College.
References
Baird,
R. C. 1968.
Aggressive
behavior
and
social
organization
in
Mollinesia
latipinna
(LeSueur).
- Tex. J.
Sci.
20: 157-
176.
Breden,
F.,
Scott, M. and
Michel, E. 1987.
Genetic
differ-
entiation
for
anti-predator
behaviour
in the
Trinidadian
guppy, Poecilia reticulate.
- Anim.
Behav. 35: 618-620.
Breden,
F and
Stoner,
G.
1987.
Male
predation
risk
determines
female
preference in
the Trinidad
guppy.
- Nature
329:
831-833.
Britton, R.
H.
and
Moser,
M. E. 1982.
Size
specific
predation
by
herons
and
its
effect on
the
sex-ratio of
natural
populations
of
the
mosquito
fish
Gambusia
affinis Baird and
Girard. -
Oecologia
53: 146-151.
Chesson,
J.
1983. The
estimation and
analysis
of
preference
and
its
relationship to
foraging
models.
- Ecology
64: 1297-
1304.
Custer,
T. W.
and
Osborn,
R.
G. 1978.
Feeding-site
description
of
three heron
species
near
Beaufort, North
Carolina.
- In:
Sprunt,
IV,
A.,
Ogden, J. C. and
Winckler, S.
(eds),
Wading
OIKOS 69:2 (1994) 257
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
birds. Research Report
7, National Audubon Society, New
York, pp. 355-360.
Endler, J. A. 1980. Natural
selection on color patterns in Poeci-
lia
reticulate. - Evolution 34: 76-91.
- 1987. Predation, light
intensity and courtship behaviour in
Poecilia reticulate
(Pisces: Poeciliidae).
- Anim. Behav. 35:
1376-1385.
Farr, J. A. 1975. The role
of predation in the evolution of social
behavior
of
natural
populations of the guppy, Poecilia retic-
ulata (Pisces:
Poeciliidae).
- Evolution 29: 151-158.
- 1989. Sexual selection
and secondary sexual differentiation
in
poeciliids:
Determinants of male mating success and the
evolution of female
choice.
- In:
Meffe, G. K. and Snelson
Jr, F.
F.
(eds),
Ecology and evolution
of
poeciliid fishes.
Prentice-Hall, New
York, pp 91-123.
- and
Travis,
J. 1986.
Fertility advertisement by female sailfin
mollies, Poecilia latipinna
(Pisces: Poeciliidae).
- Copeia
1986: 467-472.
- ,
Travis,
J.
and
Trexler,
J.
C. 1986.
Behavioural allometry
and interdemic
variation
in
sexual behaviour
of the sailfin
molly, Poecilia latipinna
(Pisces: Poeciliidae).
- Anim. Be-
hav. 34: 497-509.
Hancock,
J.
and Kushlan,
J.
1984. The
herons handbook.
-
Harper & Row, New York.
Hay,
M.
E., Renaud,
P.
E. and
Fenical,
W.
1988. Large herb-
ivores versus small
sedentary herbivores and their resistance
to
seaweed chemical defenses.
- Oecologia 75: 246-252.
Hollander,
M.
and
Wolfe,
D.
A.
1973.
Nonparametric statistical
methods.
- Wiley,
New
York.
Kramer,
D. L., Manley,
D. and
Bourgeois,
R. 1983. The effect
of
respiratory
mode
and
oxygen
concentration on the risk of
aerial
predation
in
fishes.
- Can.
J.
Zool. 61:
653-665.
Kushlan,
J. A. 1976.
Wading
bird
predation
in a seasonally
fluctuating pond.
- Auk
93:
464-476.
- 1978.
Feeding ecology
of
wading
birds.
- In:
Sprunt IV, A.,
Ogden,
J. C. and
Winckler,
S.
(eds), Wading
birds.
Research
Report 7,
National Audubon
Society,
New
York, pp.
249-
297.
- 1979. Feeding ecology
and
prey
selection
in
the white ibis.
-
Condor 81: 376-389.
Liley,
N. R. and
Seghers,
B. H. 1975. Factors
affecting
the
morphology
and
behavior of
guppies
in Trinidad.
- In:
Baerends,
G.
P., Beer,
C. and
Manning,
A.
(eds),
Function
and
evolution
in
behaviour. Clarendon
Press, Oxford, pp.
92-118.
Loftus,
W.
E and
Kushlan,
J. A. 1987.
Freshwater
fish
of
southern
Florida.
- Bull.
Fl. State Mus. Biol. Sci. 31: 147-
344.
Magurran,
A. E. and
Seghers,
B. H. 1990.
Population
differ-
ences
in the
schooling
behaviour
of
newborn
guppies,
Poeci-
lia
reticulate.
- Ethology
84:
334-342.
Manly,
B.
F J.
1985.
The statistics of natural selection
on animal
populations.
- Chapman
and
Hall,
New
York.
Ogden,
J.
C., Kushlan,
J.
A. and
Tilmant,
J. T. 1976.
Prey
selection
by
the wood stork.
- Condor 78: 324-330.
Peterson,
C. H. and
Renaud, P.
E. 1989.
Analysis of feeding
preference
experiments.
- Oecologia 80:
82-86.
Power, M.
E., Dudley,
T. L. and
Cooper, S. D. 1989. Grazing
catfish,
fishing
birds, and
attached algae in a Panamanian
stream.
- Environ.
Biol. Fishes
26: 285-294.
Reznick, D. N. 1982.
The impact of
predation
on life history
evolution in
Trinidadian guppies:
Genetic basis
of observed
life
history patterns.
- Evolution
36:
1236-1250.
- and
Endler, J. A. 1982. The
impact of
predation on life
history
evolution in
Trinidadian
guppies.
- Evolution 36:
160-177.
Roa, R. 1992. Design
and analysis
of
multiple-choice feeding-
preference experiments.
- Oecologia 89:
509-515.
Simanek, D. E. 1978.
Genetic
variability and
population struc-
ture
of
Poecilia latipinna.
- Nature 276:
612-614.
Snelson, Jr,
F F.
1984.
Seasonal
maturation and
growth of males
in a
natural
population of Poecilia
latipinna.
- Copeia 1982:
296-304.
- 1985.
Size and
morphological
variation in
males of the
sailfin
molly, Poecilia
latipinna.
- Environ. Biol.
Fishes 13:
35-47.
- and
Wetherington,
J. D. 1980. Sex ratio
in
the
sailfin
molly,
Poecilia latipinna.
- Evolution
34: 308-319.
Stoner,
G. and
Breden,
F 1988.
Phenotypic
differentiation
in
female
preference
related
to
geographic
variation in
male
predation risk
in
the
Trinidad
guppy (Poecilia
reticulate).
-
Behav.
Ecol.
Sociobiol. 22:
285-291.
Travis,
J.
1989. Ecological
genetics
of
life-history traits in
poe-
ciliid fishes.
- In:
Meffe,
G. K.
and
Snelson,
Jr,
F. F (eds),
Ecology and
evolution
of
poeciliid fishes.
Prentice-Hall,
New
York, pp.
185-200.
- and
Trexler,
J.
C. 1987.
Regional variation
in
habitat re-
quirements
of the sailfin
molly,
with
special
reference to the
Florida
Keys.
- Florida Game
and
Fresh
Water Fish
Com-
mission
Nongame
Wildlife
Program
Technical
Report
3:
1-47.
-, Farr,
J.
A.,
McManus,
M. and
Trexler,
J.
C. 1989.
Envi-
ronmental effects on adult
growth
patterns
in
the male sailfin
molly
(Poecilia
latipinna).
- Environ. Biol.
Fishes 26:
119-
127.
-, Trexler,
J. C. and
Farr,
J. A. 1993.
Body-size
variation
in
the sailfin
molly,
Poecilia
latipinna
(Pisces:
Poeciliidae):
I.
Sex-limited genetic
variation for size and
age
at maturation.
- J.
Evol.
Biol.,
in
press.
Trexler,
J. C. 1986.
Geographic
variation
in
the
sailfin
molly,
Poecilia
latipinna (Pisces:
Poeciliidae).
- Ph.
D. Thesis,
Florida State
Univ.,
Tallahassee,
FL.
- and
Travis,
J. 1990.
Phenotypic
plasticity
in the
sailfin
molly,
Poecilia
latipinna
(Pisces:
Poeciliidae):
I. Field
ex-
periment.
- Evolution
44: 143-156.
-, Travis,
J.
and
Trexler,
M.
1990.
Phenotypic
plasticity
in the
sailfin
molly,
Poecilia
latipinna
(Pisces:
Poeciliidae):
II.
Laboratory experiment.
- Evolution
44: 157-167.
-, Travis,
J.
and
McManus,
M. 1992. Effects of habitat and
body
size on
mortality
rates of Poecilia
latipinna.
- Ecology
73:
2224-2236.
258 OIKOS 69:2 (1994)
This content downloaded from 128.173.39.116 on Mon, 3 Jun 2013 11:41:40 AM
All use subject to JSTOR Terms and Conditions
