International Association for Ecology

Fluid-Mediated Dispersal in Streams: Models of Settlement from the Drift

Author(s): Dina M. Fonseca
Source: Oecologia, Vol. 121, No. 2 (1999), pp. 212-223
Published by: Springer in cooperation with International Association for Ecology
Stable URL: http://www.jstor.org/stable/4222459 .
Accessed: 14/06/2014 14:05
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 [email protected].
.
Springer and International Association for Ecology are collaborating with JSTOR to digitize, preserve and
extend access to Oecologia.
http://www.jstor.org
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
Oecologia (1999) 121:212 223 Springer-Verlag 1999
Dina M. Fonseca
Fluid-mediated
dispersal
in streams:
models of settlement from the drift
Received: 9
January
1999
/Accepted:
8 June 1999
Abstract I
propose proximal
mechanisms that
help
explain, unify,
and
expand
the
predictions
of
widely
accepted empirical
models of settlement in streams.
I
separated
the
process
that leads to settlement of a
drifting particle
into three
stages: (1)
initial contact with
a
substrate, (2) attachment,
and
(3)
settlement sensu
stricto. I used
physical principles (height
above the bed,
sinking rate,
current
speed profile)
to
predict
time until
contact
(stage 1).
I
compared
these
predictions
with
empirical
measurements of settlement of individual
black
fly
larvae
(Simulium vittatum)
in a
laboratory
flume. I
developed
models from
empirical
data for
stages
2 and 3. Each of these models is individual-based and
predicts
the fate of a
single
individual. To obtain a
population
level
prediction,
models for the three
stages
were combined and used to simulate the settlement of a
group
of black
fly
larvae. The
predictions
of this simu-
lation were
qualitatively
similar to
population
level data
from the literature
particularly
after the
incorporation
of
channel-wide
spatial heterogeneity
in current
speed.
The
effect of flow
heterogeneity
on the model
agrees
with
previous
work on the lateral
transport
of stream inver-
tebrates
during
drift events
showing
that
many organ-
isms settle
preferentially
in slower areas.
By using
D.M. Fonseca
Department
of
Biology,
University
of
Pennsylvania,
Philadelphia,
PA 19104, USA
D.M. Fonseca
(E))'
Walter Reed
Biosystematics Unit,
Museum
Support
Center
(MRC 534),
Smithsonian Institution,
4210 Silver Hill Rd.,
Suitland, MD 20746, USA
e-mail: dfonseca( erols.com,
Tel.: +
1-301-2383165, Fax: + 1-301-2383168
Present address:
'Department
of
Entomology,
Division of Communicable Diseases and
Immunology,
Walter Reed
Army
Institute of Research,
Washington
DC 20307, USA
proximal principles,
the
approach
used in this
study
brings
into focus basic
parameters
and
processes
that
influence settlement at the scale of the
organisms.
It also
provides
a null
hypothesis against
which to
study
the
effect of local flow
heterogeneity
on the settlement of
stream invertebrates and the
capacity
of
organisms
to
actively
influence settlement. Water currents in streams
and rivers
commonly transport large
numbers of
organ-
isms.
Consequently, hydrodynamic
factors that favor or
hamper
the settlement of these
organisms
can
potentially
influence distributions and abundance. Moreover,
if
settlement
probabilities vary
with flow characteristics,
this can in turn influence
foraging strategies
that
rely
on
fluid-mediated
dispersal.
Key
words Stream drift Settlement constraints
Predictive models Current
speed
Simuliidae
Intruwluction
Theoretical and
empirical
work has shown that
spatial
patchiness
and the
degree
to which
organisms
are able to
move between different
patches
can have
strong
effects
on the
dynamics
of
populations (Kareiva 1990, 1994;
Palmer and Poff
1997).
To
produce
a coherent
popula-
tion level model it is therefore
important
to characterize
habitat
heterogeneity
at scales relevant to the
organisms,
and to ascertain what factors influence movement
(i.e.,
dispersal)
at these scales.
Many organisms
reach new areas after
being
trans-
ported by
air or water currents. These include a
large
number of
aquatic
invertebrate
species,
both freshwater
and
marine,
as well as terrestrial invertebrates.
Popula-
tion models of marine intertidal and soft-bottom
sys-
tems have
emphasized
the influence of recruits from
other areas to the
dynamics
of local
populations.
This
theoretical
approach,
known as
"supply-side" ecology
(Roughgarden
et al.
1985;
Lewin 1986;
reviewed in
Grosberg
and Levitan
1992), emphasizes large-scale
movement of
organisms,
and has led to the
development
Oecologia (1999)
121:212-223
Springer-Verlag
1999
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
213
of coastal or ocean-wide
population
models
(review
in
Fevre and
Bourget 1992).
Although
these
large-scale
models
provide
valuable
information, they
fail to
explain
variations in abundance
at local
scales,
for
example
between
adjacent
areas
within the same
bay (Commito
et al.
1995).
Local
hy-
drodynamic
conditions
can, however,
be
important
de-
terminants of settlement rates
(Butman 1987;
Palmer
1988;
Butman et al.
1988;
Mullineaux and Butman
1991).
Studies of marine larvae and meiofauna have
shown that in
species
that have distinct habitat
prefer-
ences,
current
speed
can affect their
ability
to reach
preferred
habitats
(Butman
et al.
1988;
Gross et al.
1992). Except
for Palmer
(1992)
much less is known
about such settlement constraints in freshwater inverte-
brates
(see comparison
of marine and stream literature
in Downes and
Keough 1998).
Organisms
that are
transported by
water currents
constitute an
important proportion
of new colonists to
areas in a stream
(Townsend
and Hildrew
1976).
Al-
though
the factors that influence the initiation of such
fluid-mediated
dispersal (the
so-called stream
drift)
have
been studied
extensively (review
in Brittain and Eikeland
1988),
less is known about the factors that determine
when and where a
drifting organism
will return to the
substrate. This is an
important question:
the fate of
drifters
may
influence the decision of
foragers
to
disperse
via water currents
(Fonseca 1996),
and
may
affect dis-
tribution and abundance
patterns (Palmer
et al.
1996).
There is
clearly
a need to ascertain the relative
impor-
tance of
pre-
and
post-settlement processes
to the dis-
tribution of stream
organisms.
Although
this
question
has been examined
by
marine
biologists,
the information
gathered
about settlement of
marine
organisms may
not be
applicable
to settlement in
running
freshwater
systems.
Marine larvae are
quite
small and settle in
relatively
slow flows
(most
work has
been done
exposing organisms
smaller than 1 mm to
current
speeds
below 10 cm
s-~, e.g.,
Pawlik and But-
man
1993;
Turner et al.
1994).
Much of the work with
marine larvae has
emphasized
chemical clues used
by
organisms prior
to settlement
(Butman
et al.
1988).
In
contrast, many
stream invertebrates are
larger
and are
commonly exposed
to flows faster than 20 cm s-1
(Hart
et al.
1996). Although
it is not known how
important
settlement cues
may
be in
streams,
the fast flows
expe-
rienced
by
most lotic macroinvertebrates
may require
them to settle before
being
able to
recognize
the
quality
of a site
(Fonseca 1996).
Only
a small number of studies have examined the
factors that affect settlement in streams
(Walton 1978;
Ciborowski and Corkum
1979;
Otto and
Sj6str6m 1986;
Allan and Feifarek
1989;
Palmer
1992).
In the
early
1970s, however,
two
empirical
models were
published
that
predict
the distance traveled downstream
by drifting
organisms (McLay 1970;
Elliott
1971).
These models
have been used as a benchmark for the calculation of
settlement rates in streams
(Larkin
and McKone
1985).
McLay (1970) showed that the number of organisms
remaining
in
suspension
decreases
exponentially
with
distance from the area in which drift was initiated. El-
liott
(1971)
confirmed this
relationship,
and
proposed
that the distances traveled
by organisms
increased with
the
power
of current
speed.
Both these models were
developed empirically by making
measurements at the
scale of a stream channel. Measurements of current
speed
were made at mid-water
depth
in the channel.
These channel-wide
speed
measurements can
only
be
very weakly
related to the current
speed experienced by
organisms settling
on a substrate in the field
(Hart
et al.
1996).
To ascertain whether the coarse-scale
relationships
found
by
Elliott
(1971)
and
McLay (1970)
hold at finer
spatial scales,
and to understand the
processes respon-
sible for settlement
patterns,
I examined the
proximal
factors that must underlie the
empirical
results. I de-
veloped
a
simple
model from first
principles
and tested
its
predictions using
black
fly
larvae
(Simulium
vittatum
Zetterstedt, Diptera: Simuliidae).
Black
fly
larvae have
no functional
legs
and anchor themselves to silk
pads
they
attach to hard substrates
(Barr 1984). They
insert
an abdominal circlet of hooks into the silk
pad
and can
remain in the same
place
for
long periods
of time.
They
can, however,
move across hard surfaces in a
stepwise
fashion
(looping), producing
a new silk
pad
for each
step
(Reidelbach
and Kiel
1990).
Black
fly
larvae are com-
monly transported by
water currents
(Adler
et al.
1983;
Brittain and Eikeland
1988,
and references
therein),
and
while
drifting they
trail a silk thread that
may
aid set-
tlement
by adhering
to substrates
(Reidelbach
and Kiel
1990).
Previous work with S. vittatum
suggested
that set-
tlement constraints could affect the distribution of this
species.
The larvae of this
strongly
food-limited
species
(Charpentier
and Morin
1994),
should
prefer
to reside in
sites with faster
flows,
due to the
positive relationship
between
ingestion
rate
(an
important component
of fit-
ness)
and current
speed (Hart
and Latta
1986). Strong
positive relationships
between larval abundance and
current
speed
are
commonly
observed within individual
stones,
on which larvae can move via
looping (Hart
et al.
1996;
D.D. Hart and D.M.
Fonseca, personal
observations).
Between
stones, however,
where larval
dispersal
often occurs via
drifting
rather than
looping,
variations in abundance are
weakly
and
negatively
re-
lated to current
speed (D.D.
Hart and D.M.
Fonseca,
personal observations). Experiments
in which larvae
were
transplanted
between areas have also shown that
local densities do not correlate with habitat
quality,
in-
dicating
that
pre-settlement processes may
constrain
field distributions
(Fonseca 1996).
In
my
field and
laboratory
observations of black
fly
larvae,
I noticed that the
process leading
to larval
settlement
separates naturally
into three
stages
that are
affected
by
different factors.
Interestingly,
these
stages
are similar to those
proposed by
Stevenson
(1983;
personal communication)
for the settlement of diatoms
in streams. The first
stage (contact) ends when some
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
214
part
of the larva touches the substrate. The second
stage (attachment)
ends once the larva
grabs
firm hold
of the substrate
by using
its
mouthparts,
thoracic
proleg
or a silk thread.
Finally,
the third
stage,
settle-
ment sensu
stricto,
occurs when the larva makes the
first silk
pad
on the substrate. At this
point
it can
grasp
the silk
pad
with its abdominal hooks and is able to
start
feeding.
The
physical
model I describe here
predicts
the dis-
tances traveled
by
black
fly
larvae until contact with the
substrate
(stage 1). Furthermore,
I examine
empirically
the effect of current
speed
on attachment
(stage 2)
and
settlement
(stage 3)
of individual larvae.
By combining
theoretical and
empirical
information I
generate
a de-
terministic, individual-based
model,
for settlement of
black
fly
larvae in smooth flows. The models
proposed
by McLay (1970)
and Elliott
(1971)
are
population
models,
however. That
is, they
describe what
happens
when several
organisms
are released at the same time at
a
range
of
heights
above the bed.
McLay
disturbed the
streambed to force animals to
drift,
while Elliott released
groups
of
organisms
at
heights ranging
from 0 to 7 cm
above the streambed. To allow
comparisons
between the
proposed
model and the
empirical patterns
that led to
the models of
McLay (1970)
and Elliott
(1971),
I simu-
lated the settlement of a
group
of larvae released at
different
heights
above the bed and
exposed
to different
flow conditions. I restricted the
analysis
to larvae of
only
one size because Allan and Feifarek
(1989) using only
one
species
and one size class
per
trial obtained results
equivalent
to those of
McLay (1970)
and Elliott
(1971).
Taking
into consideration the observed effect of current
speed
on larval settlement I also discuss the
possibility
of
local constraints on settlement in streams and how
they
may
affect
foraging
behaviors and the distribution of
stream invertebrates.
to the substrate if no other
explanation
existed for the
removal of
organisms
from the drift. Because
predation
by
invertebrates and fish is also
possible, however,
in
McLay's experiments
R refers
only
to "the rate of dis-
appearance
from the drift". The
reciprocal
of R
(i.e., 1/R)
estimates the mean distance traveled X
by organisms
(McLay 1970).
Instead of distance one can also
express
R as a function of time from release
(number
of indi-
viduals individual-'
s-').
In that case its
reciprocal
es-
timates the mean time traveled until settlement, since
release.
Effect of current
speed
on drift distance
(Elliott 1971)
Elliott
(1971)
made similar measurements to those of
McLay (1970),
but instead of
simply disturbing
the
benthos,
he released a known number of
organisms
at
set distances
upstream
from drift nets. His results were
equivalent
to
McLay's.
The numbers of
organisms
caught
at
increasing
distances from the release
point
decreased from the
original
number
following
a
negative
exponential
curve
(Eq. 1).
Elliott
(1971) repeated
these
same
experiments
in areas of the stream with different
current
speeds.
Those data showed that the rate of dis-
appearance
of
organisms
from the drift
(R,
obtained
using Eq.
1 to fit the
data)
could be related to a
point
measure of current
speed (V,
measured at mid
channel)
by,
R =
al
V-b1
(2)
where a, and
bl
are least
square
fitted constants. The
relationship
between mean drift distance
(X
=
1/R)
and
current
speed
is
given by:
X=
(1/al)Vb' (3)
A
physical
model
Drift distance model
(McLay 1970)
McLay (1970) performed
field
experiments
in which he
disturbed the streambed in a selected area to induce
organisms
to drift. He then made collections with drift
nets at a series of distances downstream.
By
curve
fitting
the
results, McLay
came
upon
an
expression
that
pre-
dicts the number of
organisms caught
in drift nets
(Nx),
with
increasing
distance from the disturbance
(X).
That
expression
has become known as the "drift" or "drift
distance" model. It describes a
negative exponential
re-
lationship
between number of individuals
remaining
in
the drift and distance from the disturbance:
N =
Noe-RX (1)
where
No
is the number of
organisms
that
originally
entered the water
column,
and R is the
decay
rate
(number
of individuals individual-'
cm-'),
which would
reflect the rate at which
drifting organisms
were
settling
Assume an
organism
is in
suspension
in a
moving
fluid
(air or
water, Fig. 1).
Assume the
only
direction of flow
is
downstream, parallel
to a substrate. That
is,
there are
V
Ih
p . ;
1 -
[----
-
X,
Fig.
1 Schematic
depiction
of the variables included in the
physical
model that
predicts
distances traveled downstream until contact with
the bed
(Xc
the distance an
organism
will travel until
making
contact
with a substrate, h the
height
of the
organism
above the substrate,
i' its
sinking rate, V current
speed)
Models
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
215
no other directional flows
present,
like water eddies or
upward
or downward wind
gusts.
Also assume the or-
ganism
cannot swim or
fly.
Under these conditions there
are three
parameters
that will affect the distance an or-
ganism
will travel until
making
contact with a substrate
(Xc):
the
height
of the
organism
above the
substrate, h;
its
sinking rate, w; and the current
speed,
V. The
sinking
rate summarizes the effects of a
variety
of
physical
fac-
tors like
gravity,
the
viscosity
and
density
of the
water,
and the
particle's size, density,
and
shape.
The amount
of time the
organism
is in
suspension (s)
is the
height
above the substrate
(cm)
divided
by
the
sinking
rate
(cm s-l).
The distance traveled downstream
(cm)
is de-
termined
by
the time in
suspension (s) multiplied by
the
current
speed (cm s-1):
Xc
=
(h/w)
x V
(4)
This
equation
will
predict
the distances traveled
by
drifting organisms
that meet the conditions listed
above,
until
they
contact the substrate as a function of their
sinking rate,
their
height,
and current
speed.
As it
settles,
the
organism
will
experience
different current
speeds,
however.
Therefore,
V should be
interpreted
as the mean
current
speed experienced by
an
organism
until it
reaches a substrate.
Taking
that into consideration
Eq.
4
becomes:
V(x)dx
f
'=o V(x)dx
Xc
=
(h/w)* (5)
h w
V(x)
is the function that describes the
change
in current
speed
with
height (h)
above the substrate. In
simplified
flow conditions like those that can be created in a lab-
oratory flume, V(x)
will have a
logarithmic shape
as flow
slows
progressively
closer to a surface
(boundary layer,
Vogel 1994).
Methods
Experimental subjects
Sinking
rate
I calculated
sinking
rates
empirically
because the
shape
of
drifting
black
fly
larvae cannot be
represented perfectly by
a
regular geo-
metric
shape
for which there are known formulas. I measured
sinking
rate
by releasing
larvae in still water. I used a
glass
500-ml
graduated cylinder
and released
larvae,
one at a time. In
pilot
ex-
periments,
when larvae were released
just
beneath the air-water
interface it was common for some larvae to become
suspended
in
mid-water
indicating
that a silk thread
may
have become
caught
in
the surface tension. For this reason,
I released all larvae 1 cm below
the air-water interface and watched them
sinking
for 53 mm
(the
distance between the water surface and the 400 ml
mark),
before
timing
started. I then timed their fall for the next 134 mm
(until
the
100 ml
mark).
I used 22 active
larvae,
collected less than 5 h earlier
from the
field,
and 20 larvae that had been
placed
in club soda
(the
carbon dioxide renders the larvae
inert). Although
bubbles some-
times adhere to larvae that have been
placed
in club
soda,
the brief
exposure
to air
just
before
releasing
the larvae into the
experi-
mental
cylinder
eliminates them. I used least
squares regression
analysis (JMP 3.1, SAS Institute
1995)
to obtain an
equation
that
predicts
the
average sinking
rate of larvae of different sizes. I tested
the residuals for
normality using
the
Shapiro-Wilks
W-test (Sne-
decor and Cochran
1989).
I also
compared
the
sinking
rates of
active and inert larvae with a
parametric ANCOVA, using
total
length
as the
covariate, after
testing
for the
normality
of the dis-
tributions
(JMP 3.1,
SAS Institute
1995).
Laboratory
flume
I
performed experiments
in a
laboratory
flume
(Fig. 2)
constructed
following
the
design
criteria of Nowell and Jumars
(1987)
and
Denny (1988).
The
acrylic
flume was 240 cm
long
and 25.4 cm
wide. Fresh stream water was circulated
by
a
propeller (14
cm di-
ameter) powered by
a 125-W motor
(1/6 hp, Dayton 4Z528)
con-
nected to a
speed
controller
(Dayton
SCR
Control).
The
working
section was 171.5 cm
long
with water to a
depth
of 5 cm. The flume
was built to minimize flow unsteadiness:
large
vortices were broken
down
by
a fine-mesh collimator at the base of the water
tower, and
the water ascended
slowly
in the water tower
(40
x 25.4
cm,
90 cm
height)
before
gradually accelerating
into the
working
section. To
minimize flow
disturbances, the transition between the water tower
and the
working
section was done over one-third of a 30.5-cm-
diameter, 25.4-cm-wide, acrylic cylinder.
Current
speed
Black
fly
larvae
(S. vittatum)
were collected in
Taylor
Run near
Route 322 in Chester
County, Pennsylvania. They
were
placed
in
stream water in a
thermally
insulated container and taken to the
laboratory
within 1 h of
collection;
200 1 of stream water were also
brought
from the field in each occasion. In the
laboratory,
larvae
were
kept
in a circular container full of stream water re-circulated
by
a 7-cm-wide
paddle
connected to a 50-W
(1/16 hp)
motor
controlled
by
a
speed modulator,
and were used in
experiments
on
the same
day they
were collected.
During
the
experiments,
larvae were
carefully
and
gently
held
by
the narrowest area of their abdomen
using
a
pair
of size 5
forceps
and released under
water,
with their
long
axis
parallel
to the
direction of water flow and their head
facing
downstream. The
larvae can extrude
sticky
silk
through
their
mouthparts very
quickly.
This silk acted as
glue
even underwater and sometimes
prevented
larval release. When this
happened (about
once in
every
three or four
attempts)
the larvae were not used. Each larva was
only
used once.
Throughout
the
experiments
and other measure-
ments
involving
active
larvae, a
cooling
unit
kept
the water tem-
perature
within 1C of the field water
temperatures (16-19C).
Light
levels were
kept
constant
during
the
experiments
with fluo-
rescent
lights.
Current
speed
was measured with flow visualization
(Merzkirch
1974)
at different
points along
the
working
section of the flume. In
this
technique, naturally occurring neutrally buoyant particles,
like
1/6
hp
fI 2.4 m
Fig.
2
Diagram
of the main
parts
of the
laboratory
flume
used,
seen
in lateral view. The arrows
represent
the main direction of water flow.
The
working
section starts 10 cm after the
beginning
of the flat
upper
acrylic
column. The
acrylic upper
channel has a
rectangular
cross-
section. The lower channel is a PVC 15-cm-inner-diameter
pipe.
The
water column on the left is also
acrylic. Drawing
does not include
either the
angle-iron
frame or the
cooling
unit
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
216
algal
and colloidal
clumps commonly present
in stream water, were
videotaped
to measure the horizontal distances
they
moved in a
known time. I used a Panasonic WV 3240 video camera
equipped
with a 105 mm Nikkor
lens,
connected to a Panasonic NV 8950
VCR. A
digital stroboscope flashing
at 1800 strobes
per
minute was
used so that each
particle
was seen once
per
video field (each 1/30
of a
second).
The
depth
of field was
specifically
chosen to be the
shallowest
possible (f-stop
=
2.8)
to ensure that
only particles
passing
in the center of the flume were in focus. The
videotapes
were
analyzed using
a
Sony
Trinitron Color
PVM-1342Q
video
monitor. I chose
particles
at
enough heights
above the bed to
quantify
the entire
boundary layer profile.
Particles with
upward
or
downward
paths
or those out of focus were
disregarded.
A cor-
rection factor for the video
magnification
was calculated
by
mea-
suring
a ruler
videotaped immediately
before the flow visualization
run.
Using
the correction value and the
stroboscope frequency,
the
speed
of each
particle
was calculated
(cm s-').
I
plotted
the
speed
of
particles
versus their
height
above the bed
and reconstructed the
boundary layer profile
for the free stream
speed
used. To be able to
predict
current
speed
at
any height,
following
the
guidelines
of
Schlichting (1979),
I fit a
logarithmic
regression
line to the data within the first 6 mm
height
from the
bed. Visual
inspection
of the
boundary layer profile
showed that
free stream
speeds
had been reached at this
height.
Effect of
height
of release on distances traveled until contact
(stage I)
-
experimental
validation of the
physical
model
In
May 1995,
I
performed
a series of
experiments
in which black
fly
larvae
ranging
from 4.8 to 7 mm in
length
were released at a known
height
above the bed and the distances
they
traveled until con-
tacting
the flume bed were measured. The free stream current
speed
was 21.1 cm s-1. I used three
heights (2, 3,
and 4
cm)
and released
20 larvae, one at a time, from each
height.
I
performed
two trials of
ten larvae for each
height, randomizing
the order of the
height
treatments. After release,
the
point
at which each larva first made
contact with the bed was marked and the larva was collected
using
a miniature
dip
net. The distance traveled until contact was mea-
sured with a ruler to the nearest 1 mm. In these
experiments,
the
moment of first contact with the bed was
easy
to
recognize
because
at that
point
larvae
stopped
their vertical descent and started
moving
downstream in contact with the bed. The larvae were
preserved
in 80% ethanol,
and their
body length
measured to the
nearest 0.1 mm under a
dissecting microscope.
Using Eq. 5,
I
generated
the
expected
distances traveled
by
larvae of the sizes used in the
experiments
released at each
height.
After
testing
the
normality
of each distribution,
I used two-tailed
paired
t-tests
(JMP 3.1, SAS Institute
1995), blocking
for
body size,
to
compare
the observed and
predicted
results. The variance in the
predicted
data is created
by
differences in size of the larvae used.
Effect of current
speed
on distances traveled
after contact
(stages
2 and
3)
To assess the effect of current
speed
on larval attachment and
settlement,
I
performed
an
experiment
in which larvae were low-
ered in the flume until their bodies
just
contacted the bed,
after
which
they
were released. The larvae were released
facing
down-
stream with their
longitudinal
axes
parallel
to the flow because in
the
previous experiment
after the initial contact with the bed
they
would almost
invariably adopt
this
position
until
they
were able to
attach and
ultimately
settle. I chose six current
speed settings (free
stream current
speed:
12.5, 14.6, 17.5, 21.1, 23, and 27 cm
s-1),
based on the
capacity
of the larvae to settle within the
working
section of the flume. I released 20 larvae in each
speed treatment,
one at a time,
and the release was broken into two series of ten
larvae
per
current
speed
treatment. Between the two series,
the
order of the treatments was randomized.
Only
larvae
ranging
be-
tween 5.5 and 6.5 mm were used. To examine the effect of active
versus
passive
mechanisms of attachment and settlement,
six larvae
that had been
placed
in club soda were released one at a time after
the end of each active larva trial
(size range:
5.7-6.5
mm).
After each release, the larvae were observed until
they
attached
to the bed. The distance from the release
point (always
the
same)
until attachment was measured with a ruler to the nearest mm. All
larvae were collected, preserved
in 80% ethanol and their total
length
measured under a
dissecting microscope.
Larvae that failed
to attach between the release
point
and the end of the
working
section of the flume
(171.5 cm)
were not included in the least
squares regression analysis (JMP 3.1,
SAS Institute
1995).
Numerical simulations
The total distance traveled
by
a
drifting
larva before settlement is
the sum of the distances traveled downstream
during
the three
stages:
Distance until settlement
=
Distance until contact + Dis-
tance from contact until attachment + Distance from attachment
until settlement. Under all the conditions examined, larvae were
able to settle once
they
had attached. Under these conditions,
therefore, distance until settlement is
equivalent
to distance until
attachment.
Incorporating
the factors that affect contact
(Eq. 5)
and the effect of current
speed
on attachment and settlement
(ob-
tained
empirically),
the distance traveled until settlement can then
be
predicted by:
X =
(h/w)V+
cl V2
(6)
where h is the
height
above the bed at which drift was initiated, and
w is the
sinking
rate of the larvae. The variables c( and c2 were
determined
empirically.
V is the mean current
speed experienced by
a larva until
making
contact with the substrate
(equivalent
to
fh= V(x)dx/h
in
Eq. 5),
and V is free stream current
speed.
To simulate the behavior of a
group
of black
fly
larvae I fit
Eq.
6 with the
empirical
constants
appropriate
for 5.6 mm
long
S. vittatum
settling
in the
laboratory
flume. I then simulated the
relationship
between the number of larvae still in
suspension
and
distance from the release
point
under several free stream current
speeds
for a
group
of 70 larvae released at each mm from a
height
of 70 to 1 mm. I calculated the number of larvae still in
suspension
every
10 cm
starting
at 10 cm from the release
point
and
ending
when all larvae had settled. I
repeated
the calculations under five
different free stream current
speed
treatments
(12.5, 17.5, 21.1, 23,
and 27 cm
s-').
I made the same calculations for 70 larvae released
at each 1 mm from a
height
of 20 to 1 mm.
A
very conspicuous
feature of stream channels is that free-
stream current
speed
varies
laterally, being
lowest near the
margins
(Ciborowski 1983, 1987).
I included this feature in the simulation
by assuming
that after
being released, larvae in the
group experi-
enced different flow
regimes.
As a
simple approximation
to ana-
lyzing
the effects of this added
complexity,
each larva is assumed to
experience
a
single
mean flow
regime
from the time it was released
until
settling.
I
explored
the effects of more or less
heterogeneity by
contracting
or
expanding
the
range
of available current
speeds
and
placing
different
proportions
of the
group
of larvae in different flow
regimes.
Results
Relationship
between
sinking
rate and larval
length
Sinking
rate is
tightly
and
linearly
related to total
body
length (r2
=
0.955,
n =
22, F
=
1213.51,
P <
0.0001,
for active
larva).
The
longer
a larva is,
the faster it sinks.
The larvae cannot swim and while
they
are
sinking
their
bodies tend to maintain a
U-shape,
with the concave
side
facing up.
Both active larvae and larvae rendered
inert
by
carbon dioxide sank in the same
position. Very
often, however,
active larvae were observed
making
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
217
"pumping"
movements with their
upper body: they
moved their head and thoracic
proleg
back and
forth,
towards and
away
from each other. I have observed
larvae
making
the same movements
just
before
drifting
from areas of low current
speed.
I
suspect
that larvae are
extruding
silk
through
their
mouthparts
and
using
their
thoracic
proleg
to extend and form a silk thread. Be-
cause the silk is
very
thin
(averaging
80
gm,
Barr
1982)
and
virtually
invisible
underwater, however,
I was un-
able to confirm this
although
I tried several different
lighting
conditions. The
production
of silk threads
would
explain
the common occurrence of larvae released
too close to the water surface
becoming suspended
in
mid-water. Inert larvae did not move while
sinking
and
never became
suspended
in mid-water.
A
comparison
of
sinking
rates between active and
inert larvae revealed no
significant
differences
(Table 1),
indicating
no obvious behavioral control of
sinking
rate.
In the interest of
minimizing
sources of
error, however,
I
only
used data from active larvae in the least
squares
linear
regression equation relating sinking
rate to larval
length (w
= 3.362 x
length
-
1.996).
The residuals of
this linear
regression
were
normally
distributed
(Sha-
piro-Wilks
W =
0.95,
P
=
0.25).
Distances traveled downstream until contact:
comparison
of
empirical
and
predicted
values
The
empirical
and
predicted
distances traveled down-
stream until initial contact with the bed did not differ
significantly (Table 2).
The
range
of larval sizes was
narrow
(4.8-7.0 mm),
which accounts for the small
Table I To address the
possibility
of behavioral modification of
sinking rate, sinking
rates of active and inert larvae in still water
were
compared
with a
one-way
ANCOVA. The covariate is total
larval
length.
Treatment is active or inert larvae. The
adjusted
means for the two treatments are: 7.25 0.20 and 7.44
0.21 mm s
',
for active and inert
larvae, respectively
r2 = 0.97
Terms Estimates SE t-ratio P
Intercept
-1.4472 0.2944 -4.93 < 0.0001
Total
length
3.2108 0.0940 34.17 <0.0001
Treatment -0.0947 0.1473 -0.64 0.5244
Table 2
Comparison
of observed
(Obs. X)
and
predicted (Pred. X)
distance traveled downstream from a release
point,
mean
(cm)
1
SE,
until contact with the bed. Larvae were released at
three
heights: 2, 3, and 4 cm above the flume bed. The free stream
current
speed
was 21.1 cm s-1. The P-values
reported
are from
two-tailed
paired i-tests;
n
=
20 for both observed and
predicted
in
each
height
treatment
Height
Size Obs. X Pred. X P
2 0.57 0.009 29.8 1.3 27.1 0.4 0.08
3 0.56 0.011 45.1 3.8 41.3 + 0.8 0.32
4 0.57 0.007 53.2 2.4 54.9 + 1.2 0.87
scatter around the mean. I was unable to use smaller
larvae because
they
were difficult to handle without
damage.
The
average
size of larvae used in this
exper-
iment did not differ between treatment
heights
(Table 2).
Effect of current
speed
on attachment and settlement
The distance from release to attachment for active larvae
increased
significantly
with
increasing
current
speed
[distance (cm)
= 1.16 x 10-6
V57,
r2 =
0.83,
n =
76,
F =
377.82,
P <
0.0001].
I use free stream current
speed
as the
independent
variable in this
analysis
al-
though
the correct variable would be the current
speeds
within the first millimeters above the flume bed. This
practice
is warranted in these
experiments
because in
these smooth flow fields there was a
strong
correlation
between free stream current
speeds
and current
speeds
at
specific heights
above the bed. Because no larvae were
able to attach within the
working
section at the
highest
current
speed
treatment (free stream
speed
=
27 cm
s-~),
no data from this treatment were included in
the
regression analysis.
Inert larvae were unable to settle
within the
working
section of the flume in all current
speed
treatments.
At the current
speeds tested,
larvae were
usually
un-
able to attach to the bed
immediately
after release
(Fig. 3) although they
were
consistently capable
of do-
ing
so at 12.5 cm s-. At
higher
current
speeds,
larvae
slid for a while in contact with the
bed, usually
with their
head
facing
downstream. As
they slid, they
twisted and
turned, probably attempting
to
grab
onto the bed with
their
mouthparts.
At the lowest free stream
speeds
(12.5
cm s-l and 14.6 cm
s-'), they
were able to do so
E!
1 -1
I
160-
140-
120-
100-
80-
60-
40-
20-
n
0
T
0
T
1L
0
0
% 0
12 14 16 18 20 22 24 26 28
Free stream current
speed (cm s-l)
Fig.
3
Relationship
between distance traveled until attachment
(cm)
and free stream current
speed.
In each free stream
current-speed
treatment, n = 20. The distances traveled after releases at 27 cm s-'
(cross)
were 170 cm and were not included in the statistical
analysis
because no larvae were able to attach within the
working
section. The
error bars
represent
I SE
u-
I I I
w
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
218
within less than 10 cm. This
type
of attachment also
occurred often in the 17.5 cm s-1 treatment. Occasion-
ally
at 17.5 cm s-1 and
exclusively
at
higher
current
speeds, however,
larvae were
incapable
of
attaching
se-
curely
to the bed with their
mouthparts. Rather,
silk
thread
attachment,
or the attachment of silk as the
mouthparts
contacted the
bed,
was
required
for attach-
ment. The evidence for this comes from the fact that
larvae would
perform
a somersault
(i.e., turning
sud-
denly
with their head
facing upstream),
and
dangle
more
or less
briefly
in the current. Then
they
would either
pull
themselves in
using
a
technique
called
"prussiking"
in
rope-climbing (see description
in Reidelbach and Kiel
1990),
or the vortices
generated by
their
body
across the
boundary layer
would
bring
their
mouthparts
or thor-
acic
proleg
in contact with the bed. If
they
were able to
touch the bed with those
body parts they
were
eventually
able to attach.
Under the current
speed
treatments
tested,
after at-
tachment of the
mouthparts, proleg
or a silk
thread,
all
larvae were able to make a silk
pad
and insert their
abdominal hooks in
it,
after which
they
started
feeding.
Therefore,
under the conditions examined in these ex-
periments
attachment
invariably
led to settlement.
Population
level
predictions
Distance traveled until settlement
with
McLay (1970)
model
comparison
By combining
the
physical
and
empirical
information
I obtained a function that
predicts
the distance traveled
downstream until settlement
(X, cm)
for black
fly
larvae
under the conditions tested:
X
(h/(3.36L- 1.996))V
+ 1.16 x
10-6V57
where h is
height
at which drift was
initiated,
L is larval
total
length,
V is mean current
speed experienced
until
settlement,
and V is the free stream current
speed.
The simulation of the behavior of a
group
of larvae
released at different
heights
above the bed
using
this
Fig. 4A,B Hypothetical
num-
bers of larvae in
suspension
with
increasing
distance from a
release
point
after 70
larvae,
all
5.6 mm in
length (sinking
rate = 16.8 mm s-
),
were re-
leased at 1 mm increments be-
tween 1 and 70 mm above the
bed. A Predicted curves for 5
different free stream current-
speed
treatments. B Predicted
curve if a
group
of 70 larvae is
released in an area where there
are
equal frequencies
of 5 dif-
ferent free stream
speed regimes
expression
shows that the number of larvae in
suspen-
sion under each of the free stream treatments decreases
linearly
with distance from the release
point (Fig. 4A).
Most curves are truncated because under the conditions
tested,
there is
usually
a minimum distance that larvae
must drift before
settling (Fig. 4A),
even those that
started 1 mm
away
from the bed.
When larvae are released in an area with
heteroge-
neous current
speed regimes
the
relationship
between
number of larvae in
suspension
and distance from the
release
point
takes on the
negative exponential
form
predicted by McLay (1970, Fig. 4B).
To
simplify
I
only
present
the curve
generated
when five different free
stream
speed regimes
were
present
with
equal frequency
and therefore one-fifth of the larvae were
exposed
to
each current
speed regime.
The
relationship
is
expo-
nential, however, irrespective
of the actual number of
larvae
exposed
to each free stream
speed.
The effect is
strongly exponential
when
speeds
within the channel
differ
substantially, particularly
if there is a
high pro-
portion
of slow flow areas.
Effect of
current
speed
on settlement
-
comparison
with Elliott
(1971)
The
average
distance traveled downstream increases
with free stream current
speed. Figure
5 assumes that in
each free stream treatment all larvae were
exposed
to
identical flow conditions. The
range
of current
speeds
experienced
reflects
only
the
height
at which the larvae
were released. In
Fig.
5 I
compare
the curves
generated
for larvae released between 1 and 70 mm above the bed
and between 1 and 20 mm. The
relationships
between
average
distance traveled and current
speed
were fit to a
power
function
following
the
procedure
of Elliott
(1971).
In both cases distances traveled downstream
(D)
in-
crease
markedly
with free stream
speed (V)
but this effect
is
stronger
when larvae are released closer to the bed.
The
shape
of the curve for a 5.6 mm larva released 1 mm
above the bed is D = 1.5 x
10-64'9, and released 5 mm
above the bed is D = 2 x 10-3V28.
Free stream
(cm/s)
0
I-
CA
0
zo
=1
rA
4)
Cd
4-
.r.
0
0,
z
4)
Ili
0
rA
. -
0
Lt
70-
60-
50-
40-
30-
20-
10-
n
B.
0 20 40 60 80 100 120 0
20 40
20 40
0 8 1 10 140
60 80 100 120 140
Distance from release
point (cm)
I _ _X
v I I
1
Distance from release
point (cm)
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
219
I
i
l
80
1
-
1
80
i
I
60-
0
0
40-
0
~0~ ~0
20- 0
0
0
0 I I I
0
Ill
O
o
O
10 15 20 25
Free stream current speed (cm s-l)
30
Fig.
5
Hypothetical
mean distance traveled until settlement
(X)
vs.
free stream current
speed (V).
Closed circles are for larvae released
between 1 and 70 mm
height (X
=
0.8V1'3). Open
circles are for
larvae released between 1 and 20 mm
height (X
=
0.04V2)
Discussion
The first
stage
of settlement
(contact)
The
agreement
between the
predictions generated by
Eq.
5 and the observed distances traveled downstream
until contact with the bed for larvae released at
varying
heights suggests
that black
fly
larvae exerted no control
over that
stage
of the settlement
process.
For larvae in
suspension
under known
physical
conditions of current
speed
and
height
above the
bed, knowledge
of their
sinking
rate was
enough
to
predict
the distance traveled
downstream until contact.
When
measuring sinking
rate I chose to exclude
conditions under which silk threads were
caught
in the
surface tension because this was
probably
an artifact of
the
experimental technique.
I have never observed this
phenomenon
in drift events initiated
by
the larvae.
Larvae
may
be able to reduce their
sinking
rate in
flowing water, however, by releasing
a
long
silk thread
and
using
the
drag
on the silk to
carry
them down-
stream. I have observed larvae
being transported by
water currents in current
speeds
under which inert larvae
remain immobile on the bed. Because the silk is
virtually
invisible under
water,
I have not been able to confirm
this. If
confirmed,
this behavior would be akin to that
used
by ballooning spiders (Suter 1991; Weyman 1993),
Lepidoptera
larvae
(Leonard 1971;
McManus and Ma-
son
1983;
Cox and Potter
1986)
and some marine bi-
valves
(Beukema
and Vlas
1989;
Martel
1993;
Martel
and Diefenbach
1993).
Larval behavior
during
attachment and settlement
Although
larvae acted as
passive particles
until
making
contact with a substrate
(stage 1), they
exerted control
over the second and third
stages (attachment
and set-
tlement, respectively).
Larvae were able to attach to the
bed and
subsequently
settle under flow
regimes
where
inert larvae were
swept away.
Attachment was
strongly
influenced
by
current
speed,
however.
High
current
speed
increased the distances traveled in contact with the
bed before attachment.
As
mentioned,
I used free stream current
speed
as a
proxy
for the current
speeds experienced by
larvae while
drifting
in contact with the flume bed. I was able to do
so,
because the flow fields used in these
experiments
were
smooth and the
boundary layer
was well
developed.
However,
this
may
not be true over natural substrates in
the field
(Hart
et al.
1996)
which is one of the reasons
why
teasing apart
some of the
processes
behind settlement
from the drift
may
be more feasible in
laboratory
flumes.
During
this
study
in which water turbulence was
minimized and current
speeds
were
relatively slow,
suc-
cessful attachment
always
led to settlement. Under faster
current
speeds
or
very
intense turbulence
conditions,
however,
this
may
not
always
be the case. A
particularly
sensitive time is when larvae have
just
attached a silk
thread to the bed but have not been able to attach a silk
pad. Particularly
at the
highest
current
speeds
tested
there were occasions in which larvae
dangled
in the
current like a kite. In the
field,
I have observed larvae
reaching
this
stage
and not
settling (so
have Reidelbach
and Kiel
1990).
It is
possible
the larvae
may
have de-
cided not to settle, but it is also conceivable that the silk
thread
may
break due to the sudden accelerations as-
sociated with the turbulent flow
regimes
that are com-
mon in streams
(Hart
et al.
1996).
Local current
speed
decreases
probability
of settlement
The results of this
study predict
that
drifting may
not
always
be a
good strategy
for
reaching
areas with
high
current
speeds,
at least for
organisms
with limited
ability
to swim. The distances traveled before settlement in-
crease
exponentially
with current
speed
so the
proba-
bility
of settlement on a substrate of finite size decreases
rapidly
with
increasing
current
speed
above that sub-
strate.
Despite
the
negative
effect of current
speed
on set-
tlement, however, black
fly
larvae occur at
high
densities
in
fast-flowing
areas in streams
(riffles)
and are almost
absent from
very
slow areas
(pools).
The
positive
cor-
relation between larval abundance and flow at this
"riffle-pool"
scale
(D.D.
Hart and D.M.
Fonseca, per-
sonal
observations)
can be
explained by
the fact that
black
fly
larvae will drift
readily
from slow flow areas
(Fonseca
and Hart
1996). Leaving
an area
by
drift in
search of a better
feeding
site
may
be
advantageous
for
larvae in
pools
because larvae
drifting
from
pools
into
riffles
may
have a
high probability
of
settling
in a faster
area than the one
they just
left. Once the
organism
ex-
periences
a current
speed
above a
possibly species spe-
cific
threshold, however,
the
organism may
be
unwilling
to drift because of the expected diminishing returns
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
220
(Fonseca 1996).
This would lead to the
acceptance
of
sub-optimal feeding
areas and to a lack of correlation
between larval
density
and current
speed.
Such lack of
correlation has been
interpreted
as
showing
that current
speed
is
unimportant
to black
fly
larval survival
(Eym-
ann
1993).
The fact that larval distributions are
tightly
and
positively
correlated with current
speed
when mea-
surements are made within a
single
substrate
(Hart
et al.
1996;
D.D. Hart and D.M.
Fonseca, personal
observa-
tions) clearly
underscores the need to ascertain the im-
portance
of settlement constraints in
generating patterns
of distribution.
It is then at intermediate
scales, mainly
across sub-
strates within areas of moderate to
high
current
speeds,
that the
negative
effect of current
speed
on settlement
may
affect
foraging decisions,
distributions and abun-
dance. At that
scale, oviposition by
adult females
may
be
an
important
source of larvae to areas in fast flows. In
S.
vittatum, egg
mass
density
is
highly
and
positively
correlated to current
speed (D.D.
Hart and D.M. Fon-
seca, personal observations)
and first instars will remain
on or near the
oviposition
site unless flows are slow
and/
or
conspecific
densities are too
high (Fonseca
and Hart
1996).
There are also some indications
(Fonseca 1996)
that features of the stream
bed,
like rocks
breaking
the
water surface, may
create
high
settlement conditions
either
by generating
slow flow areas downstream from
them,
or
by producing
flow
patterns
that increase the
probability
of contact
by drifting organisms.
It is im-
portant
to
emphasize
that
although they
are
legless,
black
fly
larvae can
loop
over hard substrates
relatively
fast
(Reidelbach
and Kiel
1990).
Due to the
high
het-
erogeneity
of flows over a
single
substrate
(Hart
et al.
1996),
larvae can
conceivably
settle from the drift in
slower flows if those are
present,
and then
loop
to faster
flows within the same substrate. The
importance
of these
and
possibly
other factors for distributions in streams
needs to be addressed before a full
predictive
model can
be
designed.
From the
laboratory
to the field
The work described in this
paper
constitutes a first
step
towards
understanding
the
processes influencing
the
settlement of stream macroinvertebrates. It should be
emphasized
that the flow
regimes
to which black
fly
larvae were
exposed
in these
laboratory experiments
represent
a limited subset of the conditions
commonly
experienced by settling
larvae. In these
experiments
flow
unsteadiness was
kept
to a minimum. Some
preliminary
results demonstrate that turbulence increases
settlement,
however under all the conditions tested turbulent flows
did not cancel the
negative
effect of current
speed (D.M.
Fonseca, personal observations).
Although
the ultimate aim is to
predict
the fate of
drifting organisms
in the field, there is a clear need to
initially
divide the
process
into its various
physical
and
behavioral elements, and then test each of these in sim-
plified
conditions.
Subsequently,
different
types
of com-
plexity
can be added and its effects examined. The
goal
is
ultimately
to
replicate
field conditions. At this
point,
however,
there are
already
some
predictions
that
appear
robust. For
example: (1)
local current
speed
has a
neg-
ative effect on the
probability
of
contact; (2)
larval size
correlates with
probability
of contact with the substrate
through
its effect on
sinking rate; (3) height
of release
affects the distance traveled until settlement. Further
work needs to be done to examine how flow unsteadiness
and behavioral control affect these and other
predic-
tions. In
particular,
the
processes influencing
attachment
and settlement sensu stricto need to be elucidated.
Comparison
of
predictive
models of settlement
After a threshold distance below which no larvae is
capable
of
settling (a phenomenon
also observed
by
Elliott
1971),
the numerical simulation
predicts
that the
relationship
between number of larvae in
suspension
and
distance from the release
point
is linear
(Fig. 4A).
This
differs from the
negative exponential relationship
ob-
served
by McLay (1970).
This
prediction
will
only hold,
however,
if all larvae are released in identical flow con-
ditions. A simulation that includes
spatial
variation in
current
speed predicts
that the
relationship
between
number of larvae
remaining
in the water column and
distance from the release
point
will have the familiar
exponential shape (Eq. 1, Fig. 4B).
It is conceivable that
other
processes may
lead to the same results. Never-
theless,
it is well known that vertical
speed profiles vary
laterally
across stream channels
(Ciborowski 1983)
and
it is certain that all-channel
experiments expose organ-
isms to a
variety
of flow
regimes.
Therefore,
incorpo-
rating spatial
variance in current
speed
is not unrealistic
and it is
heartening
that it was sufficient to match em-
pirical
data.
Analysis
of this simulation reveals that the
expo-
nential
shape
results from the initial
settling
of
organ-
isms in areas of slow
flow,
while those
experiencing
higher
flows continue in
suspension.
This not
only agrees
with
reports
that
drifting organisms
are
commonly
car-
ried to areas of slow flow near the
margins (Ciborowski
1983),
but also underlines the need to measure flow at
local
scales,
and the
importance
of current
speed
as a
deterrent to settlement.
When
compared
with the
findings
of Elliott
(1971)
the
predictions
of the model are also
revealing.
The re-
lationship
between mean distance and current
speed
can
be described
by
a
power
function
(Eq. 3, Fig. 5).
But
while in the
experiments
of Elliott
(1971) b1
= 1.3 for
Simulium
spp.,
in the simulation the
exponent bl
varies
with the
height
at which larvae were released. The
higher
the
average height-of-release
the smaller the
exponent.
This result makes intuitive sense: if
organisms
are re-
leased
high
in the water column, the distances traveled
until contact
(which
increase
linearly
with current
speed,
bl
=
1, Eq. 5),
will be so
large they mask, in a statistical
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
221
sense,
the effect of current
speed during
the next
stages
of settlement. This
may
be the reason the
relationship
between distance and current
speed
is often linear in
conventional,
all-channel
experiments (Elliott 1971;
Ciborowski and Corkum
1979;
Allan and Feifarek
1989;
Larkin and McKone
1985).
The value of the
exponent,
therefore, provides
no information about the
probability
of settlement in different local flow conditions. This
finding again
underscores the
inability
of
experiments
performed
at
large
scales to
capture
all the
processes
that
affect settlement of
organisms
in
heterogeneous
envi-
ronments.
In
conclusion,
simulations
using
the model
presented
in this
paper
can match
empirical
data from the litera-
ture. The conditions under which the model fits
empir-
ical data are
quite parsimonious
and
may pinpoint
mechanisms that underlie the
patterns
found in nature.
The model
presented highlights
the
shortcomings
and
also
provides
a mechanistic
bridge
between the two
pre-
existing
models.
Although
it was obvious that the
models of both
McLay (1970)
and Elliott
(1971)
had to
stem from the same
group
of
processes
the connections
had not
yet
been
formally proposed.
Do local current
speeds
influence distributions
in streams?
Although
current
speed
decreases the
probability
of
settlement of an individual larva on a
substrate,
at the
population
level the number of
organisms drifting
over a substrate
per
unit time increases with current
speed.
The flux of
organisms (number
of individuals
cm-1 x cm s-1
=
number of individuals
s-1, F)
should
increase
linearly
with current
speed (F
=
aV, Fig. 6A).
Fig.
6A-E The number of col-
onists is influenced both
by
the A.
rate of
delivery
of
organisms
and
by
the rate of settlement. If
A flux
(number
of individu-
als
s-')
increases
linearly
with
current
speed,
and B settlement
/
rate
(number
of individu-
als individual-'
s-1)
increases
with the inverse of current
speed,
then C the number of
colonists
(number
of individu-
als)
will be
independent
of
Current
speed
current
speed. However,
if
D the
relationship
between set-
tlement and current
speed
has
an
exponent
smaller than -1,
for
example -1.5, then E num-
ber of colonists will decrease
with current
speed
Therefore current
speed
has two
opposite
effects at the
population
level: decreased settlement but increased
availability
of
organisms.
Because Elliott
(1971),
and
more
recently
Allan and Feifarek
(1989), report
several
cases in which rates of settlement
(number
of individu-
als individual-'
s-~)
increased with the inverse of cur-
rent
speed (R
= b ',
Fig. 6B),
it has been
argued
that
the two effects of current
speed
cancel each other. In that
case the final number of colonists
(number
of individu-
als),
is
independent
of the local current
speed (Fig. 6D).
Hinging
on a
strictly
linear inverse
relationship
be-
tween settlement rate and current
speed,
this
argument
is
weak for several reasons.
First,
the fact that
organisms
are
encountering
a
variety
of local flow
regimes upon
reaching
the streambed adds variance to the data. That
variance decreases the statistical
power
of the curve-fit-
ting
tests. There are several studies in which
exponents
smaller than -1 were found in the
relationship
between
settlement rate and free stream current
speed (Elliott
1971;
Ciborowski and Corkum
1979;
Allan and Feifarek
1989).
Those
exponents are, however,
not
statistically
different from -1 due to
high
variance
(D.M. Fonseca,
personal observations). Second,
as I have mentioned
before,
when
organisms
are released
away
from the
streambed,
the
resulting
rate of settlement reflect over-
whelmingly
the distances traveled
by drifting organisms
until
making
contact with a substrate. These distances
increase
linearly
with current
speed,
as is clear from
analysis
of the
physical
model
(Eq. 5).
The
magnitude
of
the effects of current
speed
at this scale therefore de-
creases the
probability
of detection of constraints on
settlement at local scales.
Finally,
it is also
possible
the
relationship
between flux of
organisms
and current
speed may
not be linear at local scales. This could occur
for
example,
if
organisms
have a
higher tendency
to exit
B. C.
z Z
Current
speed
Current
speed
D. E.
a
Current
speed
Current
speed
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
222
slow than fast areas
(Corkum
et al.
1977;
Fonseca and
Hart
1996), providing
a
high
number of colonists to
other slow areas close
by.
In the
laboratory experiments reported
in this
paper
the rate of settlement of black
fly
larvae decreases with
the
power
of current
speed
with
exponents
smaller than
-1
(Fig. 6D).
In this
case,
the
putative
linear increase in
flux of
organisms
with current
speed
does not cancel the
negative
effect of current
speed (Fig. 6E). Although
one
may argue
that black
fly
larvae are
particularly
vulner-
able to settlement constraints because
they
cannot
swim,
they
have a
sophisticated anchoring system
that allows
them to attach to substrates even in
very
fast flows
(Reidelbach
and Kiel
1990).
In
contrast,
Elliott
(1971),
for
example, reported seeing nymphs
of
Rhithrogena
semicolorata,
a
mayfly usually
found in
fast-flowing
waters, touching
the substratum several
times,
while
traveling downstream,
before
being
able to settle.
Also,
Ciborowski
(1987)
found that
mayfly nymphs
that usu-
ally
occur in fast flows are
transported laterally
to slow
flow areas in the
margins
of the stream after
drifting.
It
is
possible, therefore,
that
although swimming organ-
isms
may
contact the substrate faster than
predicted by
the
physical model, they may
have
difficulty attaching
and
settling.
Fast-flowing
areas have
high
rates of food
delivery
that favor
suspension-feeders,
and
high algal production
(Pfeiffer
and McDiffett
1975)
that favors
grazers.
Al-
though
in streams it is estimated that
up
to 80% of the
organisms colonizing
new substrates are
transported by
water currents
(Brittain
and Eikeland 1988 and refer-
ences
therein; Mackay 1992),
the
possibility
of the ex-
istence of constraints to the settlement of lotic
organisms
has received almost no attention. I
propose
that con-
straints to settlement from the drift onto fast
flowing
areas
may impact
the
foraging
behavior and distribu-
tions of stream
organisms
and
ultimately population
dynamics
in streams.
Acknowledgements
I thank Warren Ewens,
Erik Silldorff, and Leo
Shapiro
for
very
fruitful discussions. I thank also David Hart,
Margaret Palmer, Jim McNair, Peter Petraitis, Arthur Dunham,
Leo
Shapiro,
Steve Kohler and Barbara Downes for
providing
constructive comments on earlier versions of this
manuscript.
This
paper
includes research
reported
in a doctoral dissertation com-
pleted
at the
University
of
Pennsylvania
and was
supported by
NSF Dissertation
Improvement
Grant DEB-9224214.
Referencus
Adler PH, Light RW,
Kim KC
(1983)
The
aquatic
drift
patterns
of
black flies
(Diptera: Simuliidae). Hydrobiologia
107:183-191
Allan JD, Feifarek BP
(1989)
Distances traveled
by drifting mayfly
nymphs:
factors
influencing
return to the substrate. J N Am
Benthol Soc 8:322-330
Barr WB
(1982)
Attachment and silk of larvae of Simulium vittatum
Zetterstedt
(Diptera: Simuliidae).
Masters thesis, University
of
Alberta, Canada
Barr WB
(1984) Prolegs
and attachment of Simulium vittatum
(sibling IS-7) (Diptera: Simuliidae)
larvae. Can J Zool 62:1355-
1362
Beukema JJ, Vlas JD
(1989)
Tidal-current
transport
of thread-
drifting postlarval juveniles
of the bivalve Macoma balthica from
the Wadden Sea to the North Sea. Mar Ecol
Prog
Ser 52:193-200
Brittain JE, Eikeland TJ
(1988)
Invertebrate drift
-
a review.
Hydrobiologia
166:77-93
Butman CA
(1987)
Larval settlement of soft-sediment inverte-
brates: the
spatial
scales of
pattern explained by
active habitat
selection and the
emerging
role of
hydrodynamical processes.
Oceanogr
Mar Biol Annu Rev 25:113-165
Butman CA, Grassle
JP,
Webb CM
(1988)
Substrate choices made
by
marine larvae
settling
in still water and in a flume flow.
Nature 333:771-773
Charpentier B, Morin A
(1994)
Effect of current
velocity
on
inges-
tion rates of black
fly
larvae. Can J Fish
Aquat
Sci 51:1615-1619
Ciborowski JJH
(1983)
Downstream and lateral
transport
of
nymphs
of
two-mayfly species (Ephemeroptera).
Can J Fish
Aquat
Sci 40:2025-2029
Ciborowski JJH
(1987) Dynamics
of drift and microdistribution of
two
mayflies populations:
a
predictive
model. Can J Fish
Aquat
Sci 44:832-845
Ciborowski JJH, Corkum LD
(1979) Importance
of behaviour to
the re-establishment of
drifting Ephemeroptera.
In:
Flannagan
JF,
Marshall KE
(eds)
Advances in
Ephemeroptera biology.
Plenum, New
York, pp
321-330
Commito JA, Currier CA, Kane LR
(1995) Dispersal dynamics
of
the bivalve Gemma
gemma
in a
patchy
environment. Ecol
Monogr
65:1-20
Corkum LD, Pointing PJ, Ciborowski JJH
(1977)
The influence of
current
velocity
and substrate on the distribution and drift of
two
species
of
mayflies (Ephemeroptera).
Can J Zool 55:1970-
1977
Cox DL, Potter DL
(1986)
Aerial
dispersal
behavior of larval
bagworms, Thyridopteryx ephemeraeformis (Lepidoptera:
Psychidae).
Can Entomol 118:525-536
Denny
MW
(1988) Biology
and the mechanisms of the
wave-swept
environment. Princeton
University Press, Princeton
Downes BJ, Keough
MJ
(1998) Scaling
of colonization
processes
in
streams:
parallels
and lessons from marine hard substrata. Aust
J Ecol 23:8-26
Elliott JM
(1971)
The distances traveled
by drifting
invertebrates in
a Lake District stream.
Oecologia
6:350-379
Eymann
M
(1993)
Some
boundary layer
characteristics of micro-
habitats
occupied by
larval black flies
(Diptera: Simuliidae).
Hydrobiologia
259:57-67
Fevre JL, Bourget
E
(1992) Hydrodynamics
and behaviour:
transport processes
in marine invertebrate larvae. Trends Ecol
Evol 7:288-289
Fonseca DM
(1996)
Fluid-mediated
dispersal:
effects on the for-
aging
behavior and distribution of stream insects. PhD disser-
tation, University
of
Pennsylvania, Philadelphia
Fonseca DM, Hart DD
(1996) Density-dependent dispersal
of
black
fly
neonates is mediated
by
flow. Oikos 75:49-58
Grosberg
RK, Levitan DR
(1992)
For adults
only? Supply-side
ecology
and the
history
of larval
biology.
Trends Ecol Evol
7:130-133
Gross TE, Werner FE, Eckman JE
(1992)
Numerical
modeling
of
larval settlement in turbulent bottom
boundary layers.
J Mar
Res 50:611-642
Hart DD, Latta SC
(1986)
Determinants of
ingestion
rates in filter-
feeding
larval blackflies
(Diptera: Simuliidae).
Freshwater Biol
16:1-14
Hart DD, Clark BD, Jasentuliyana
A
(1996)
Fine-scale field mea-
surement of benthic flow environments inhabited
by
stream
invertebrates. Limnol
Oceanogr
41:297-308
Kareiva P
(1990) Population dynamics
in
spatially complex
envi-
ronments:
theory
and data. Philos Trans R Soc Lond B
330:175-190
Kareiva P
(1994) Space:
the final frontier for
ecological theory.
Ecology
75:1
Larkin PA, McKone DW
(1985)
An evaluation
by
field
experi-
ments of the
McLay
model of stream drift. Can J Fish
Aquat
Sci 42:909-918
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions
223
Leonard DE
(1971)
Air-borne
dispersal
of larvae of the
gypsy
moth
and its influence on
concepts
of control. J Econ Entomol
63:638-641
Lewin R
(1986) Supply-side ecology.
Science 234:25-27
Mackay
RJ
(1992)
Colonization
by
lotic macroinvertebrates: a
review of
processes
and
patterns.
Can J Fish
Aquat
Sci 49:617-
628
Martel A
(1993) Dispersal
and recruitment of zebra mussel
(Dreissena polymorpha)
in a nearshore area in west-central lake
Erie: the
significance
of
postmetamorphic drifting.
Can J Fish
Aquat
Sci 50:3-12
Martel
A, Diefenbach T
(1993)
Effects of
body size,
water current
and microhabitat on mucous-thread
drifting
in
post-metamor-
phic gastropods
Lacuna
spp.
Mar Ecol
Prog
Ser 99:215-220
McLay
C
(1970)
A
theory concerning
the distance traveled
by
animals
entering
the drift of a stream. J Fish Res Bd Can
27:359-370
McManus
ML,
Mason CJ
(1983)
Determination of the
settling
velocity
and its
significance
to larval
dispersal
of the
gypsy
moth
(Lepidoptera: Lymantriidae).
Environ Entomol 12:270-272
Merzkirch W
(1974)
Flow visualization. Academic
Press,
New
York
Mullineaux
LS,
Butman CA
(1991)
Initial
contact, exploration
and
attachment of barnacle
(Balanus amphitrite) cyprids settling
in
flow. Mar Biol 110:93-103
Nowell
ARM, Jumars PA
(1987)
Flumes: theoretical and
experi-
mental considerations for simulation of benthic environments.
Oceanogr
Mar Biol Annu Rev 25:91-112
Otto C, Sjostrom
P
(1986)
Behaviour of
drifting
insect larvae.
Hydrobiologia
131:77-86
Palmer MA
(1988) Epibenthic predators
and marine meiofauna:
separating predation, disturbance,
and
hydrodynamic
effects.
Ecology
69:1251-1259
Palmer MA
(1992) Incorporating
lotic meiofauna into our under-
standing
of faunal
transport processes.
Limnol
Oceanogr
37:329-341
Palmer
MA,
Poff NL
(1997)
The influence of environmental het-
erogeneity
on
patterns
and
processes
in streams. J N Am
Benthol Soc 16:169-173
Palmer
MA, Allan
JD,
Butman CA
(1996) Dispersal
as a
regional
process affecting
the local
dynamics
of marine and stream
benthic invertebrates. Trends Ecol Evol 11:322-326
Pfeiffer
RJ,
McDiffett WF
(1975)
Some factors
affecting primary
productivity
in stream riffle communities. Arch
Hydrobiol
75:306-317
Pawlik
JR,
Butman CA
(1993)
Settlement of a marine tube worm
as a function of current
velocity: interacting
effects of
hydro-
dynamics
and behavior. Limnol
Oceanogr
38:1730-1740
Reidelbach
J,
Kiel E
(1990)
Observations on the behavioral se-
quences
of
looping
and
drifting by blackfly
larvae
(Diptera:
Simuliidae). Aquat
Insects 12:49-60
Roughgarden J, Iwasa
Y,
Baxter C
(1985) Demographic theory
for
an
open
marine
population
with
space-limited
recruitment.
Ecology
66:54-67
SAS Institute
(1995)
JMP 3.1. SAS
Institute, Cary
Schlichting
H
(1979) Boundary layer theory,
7th edn. McGraw-
Hill,
New York
Snedecor
GW, Cochran WG
(1989)
Statistical methods. Iowa State
University Press, Ames
Stevenson RJ
(1983)
Effects of current and conditions
simulating
autogenetically changing
microhabitats on benthic diatom im-
migration. Ecology
64:1514-1524
Suter RB
(1991) Ballooning
in
spiders:
results of wind tunnel ex-
periments.
Ethol Ecol Evol 3:13-25
Townsend
CR,
Hildrew AG
(1976)
Field
experiments
on the
drifting,
colonization and continuous redistribution of stream
benthos. J Anim Ecol 45:759-772
Turner
EJ, Zimmer-Faust
RK,
Palmer
MA, Luckenbach
M,
Pentcheff ND
(1994)
Settlement of
oyster (Crassostrea virginica)
larvae: effects of water flow and a water-soluble chemical cue.
Limnol
Oceanogr
39:1579-1593
Vogel
S
(1994)
Life in
moving
fluids. The
physical biology
of
flow,
2nd edn. Willard
Grant,
Boston
Walton OE Jr
(1978)
Substrate attachment
by drifting aquatic
in-
sect larvae.
Ecology
59:1023-1030
Weyman
GS
(1993)
A review of the
possible
causative factors and
significance
of
ballooning
in
spiders.
Ethol Ecol Evol 5:279-
291
This content downloaded from 165.123.34.86 on Sat, 14 Jun 2014 14:05:17 PM
All use subject to JSTOR Terms and Conditions