Medolian C. Pressley. THE ROLE OF DENSITY, PREDATOR DEFENSE, AND
MICROHABITAT PREFERENCE ON THE SPATIAL DISTRIBUTION OF
Chromopleustes oculatus (CRUSTACEA: AMPHIPODA) AT BELL ISLAND,
WASHINGTON. (Under the direction of Lisa M. Clough and Roger A. Rulifson)
Department of Biology, June 2003.
This study examined abiotic and biotic effects on the spatial distribution of
Chromopleustes oculatus during mid-May to late-July 2000. In situ video ofA.
fimbriatum recorded spatial distributions of the amphipods in the field at high and low
slack tide. Testing pools were designed to isolate the roles of density and the presence of
predators on spatial distributions. I assessed habitat use by presenting amphipods with
five substrates: sea cucumbers, rocks, Agarum blades, sand and shell mixtures, and sand.
Amphipod numbers were greatest at the base oiA. fimbriatum (20 individuals per image)
and smallest at the tip (eight individuals per image). Mean nearest neighbor distances
(MNND) significantly decreased from tip to base. MNND for four densities in testing
pools (500, 1000, 2000, and 5000 individuals-0.664 m'^) significantly decreased as
density increased from 500 to 2000 individuals. MNND in pools containing 5000 and
1000 individuals were not statistically different. The MNND in the presence of a naïve
fish predator, Oligocottus maculosus, was higher than the MNND in the predator’s
absence. Amphipods occupied the highest density per area on sea cucumbers
(Cucumaria miniata). Amphipods attempting to mate in testing pools, in addition to
possible cannibalism, were most likely the result for the increased MNND in density and
predator experiments. Multiple abiotic and biotic factors influenced the spatial
distribution of C. oculatus, however, this study would argue that chemical defense does
not play an important role for C. oculatus with respect to amphipod position on algal
blades, habitat choices, predators, reproduction, and in some cases cannibalism.
THE ROLE OF DENSITY, PREDATOR DEFENSE, AND MICROHABITAT
PREFERENCE ON THE SPATIAL DISTRIBUTION OF Chromopleustes oculatus
(CRUSTACEA: AMPHIPODA) AT BELL ISLAND, WASHINGTON
A Thesis
Presented to
the Faculty of the Department ofBiology
East Carolina University
In Partial Fulfullment
of the Requirements for the Degree
Master of Science in Biology
Medolian C. Pressley
July 2003
THE ROLE OF DENSITY, PREDATOR DEFENSE, AND MICROHABITAT
PREFERENCE ON THE SPATIAL DISTRIBUTION OF Chromopleustes oculatus
(CRUSTACEA: AMPHIPODA) AT BELL ISLAND, WASHINGTON
By
Medolian C. Pressley
APPROVED BY:
DIRECTOR OF THESIS
Lisa M. Clough, Ph.D.
DIRECTOR OF THESIS
Roger A. Rulifson, Ph.D.
COMMITTEE MEMBER
CHAIR OF THE DEPARTMENT OF BIOLOGY
Ronald J. Newton, Ph.D
INTERIM DEAN OF THE GRADUATE SCHOOL
V
Paul D. Tschetter, Ph.D.
ACKNOWLEDGMENTS
I thank Dr. Lisa M. Clough and Dr. Roger A. Rulifson, both ofwhom agreed to
take responsibility to lead this thesis into its final stages. Additionally, I thank them for
their abundant time in their busy schedules and their constructive criticisms in revisions.
Dr. Rulifson and Dr. Anthony Overton provided substantial help in final editorial
revisions to meet Graduate School criteria. I thank Dr. Gerhard W. Kalmus for his
continued support in all the stages of this thesis and to the Biology Department for its
support. I give special thanks to Marcy Hutchinson for her diving companionship and
overall support at both Friday Harbor, Washington, and Greenville, North Carolina.
Without her help, none of the collections and field pictures would have been possible. I
thank Friday Harbor Laboratories, University ofWashington Diving Safety Office, and
East Carolina University Diving and Water Safety for use of their facilities, boats, and
underwater equipment. Thanks to Tim Charles for SEM pictures of Chromopleustes
oculatus. I wish to thank my committee members for their patience and criticisms in the
text: Dr. Richard D. Hauser, Dr. Kyle D. Summers, and Dr. Terry L. West. I also thank
Dr. Stephen F. Norton for his ideas and insights in creating the research design. This
study was funded in part by grants from the PADI Foundation, Sigma Xi - The Scientific
Research Society, the Department of Biology, and from personal funds.
TABLE OF CONTENTS
LIST OF TABLES v
LIST OF FIGURES vii
INTRODUCTION 1
Abiotic Factors 1
Biotic Factors 2
HYPOTHESES 18
Distributions and Patterns 18
Density 19
METHODS 20
Study Site 20
Standardization 20
Nearest Neighbor Calculations 21
Field Observations 24
Laboratory Experiments 28
Spatial distributions and patterns of aggregates 29
Predators and spatial distributions 30
Habitat choice experiments 31
RESULTS 35
Standardization 35
Field Observations 35
Laboratory Experiments 53
Spatial distributions and patterns of aggregates 56
Predators and spatial distributions 63
Habitat choice experiments 68
DISCUSSION 72
Standardization 72
Field Observations 72
Spatial distributions 74
Density 74
Laboratory Experiments 77
Spatial distributions and patterns of aggregates 77
Predators and spatial distributions 81
Habitat choice experiments 82
CONCLUSIONS 87
LITERATURE CITED 89
LIST OF TABLES
1. Percentage of nearest neighbor distances < 0.5 pixels ( <0.56 cm for field
observations and < 0.22 cm for laboratory manipulations) that were
eliminated from nearest neighbor distance calculations due to possible
mating or cannibalism by Chromopleustes oculatus 23
2. Amphipod variability, expressed as Mean + SE, from counts of 12 algal
blades taken as a field survey in situ at Bell Island 13 June 2000 through 25
June 2000 during high and low slack tide prior to beginning field
observations 38
3. ANOVA of the differences in mean number of individuals at three positions
on Agarum fimbriatum (base, center, and tip) using images taken at high
and low slack tides ofnatural populations of Chromopleustes oculatus at
Bell Island 44
4. Bonferroni Post Hoc test comparing the mean abundances of
Chromopleustes oculatus at three locations on Agarum fimbriatum', base,
center, and tip 46
5. ANOVA of the differences in mean nearest neighbor distances at three
positions on Agarum fimbriatum (base, center, and tip) using images taken
at high and low slack tides ofnatural populations of Chromopleustes
oculatus at Bell Island 54
6. Bonferroni Post Hoc interaction results for mean nearest neighbor distances
that shifted as a function of tidal state (High / Low Slack tide) at a given
blade position (Base / Center / Tip) for Chromopleustes oculatus 55
7. ANOVA of the relative change in mean nearest neighbor distances of
Chromopleustes oculatus at 2 and 24 hours in experiments manipulating
four densities (500, 1000, 2000, and 5000 individuals per 0.664m^) 58
8. Bonferroni Post Hoc comparisons ofmean nearest neighbor distances
analyzed from images ofChromopleustes oculatus taken at five preset
locations (north, south, east, west, and center) in testing pools for four
densities, respectively (500, 1000, 2000, and 5000 individuals 0.664m^) 58
9. Bonferroni Post Hoc interaction results for mean nearest neighbor distances
that shifted as a function of density (500, 1000, 2000, and 5000 individuals
0.664m ) at a given time (2 hr and 24 hr) for Chromopleustes oculatus
(comparisons are sorted by increasing probabilities) 64
10. Table of values used in Chi-Square analyses; two-dimensional percentage
of cover for each habitat in aquaria, expected number of Chromopleustes
oculatus with respect to habitat percentages, and observed numbers of
individuals in aquaria
VI
LIST OF FIGURES
1. Map of San Juan Islands National Wildlife Refuge at Friday Harbor
Laboratories in the state ofWashington, USA 8
2. Patterns of distribution for Chromopleustes oculatus from photoquadrats
taken summer 1999 based on an Index ofAggregation (R) 12
3. Detailed map featuring Bell Island, the primary field site, located south of
Oreas Island and north of Shaw Island 14
4. Image of the Colander Weed, AgarumJimbriatum, the dominant spatial
habitat at Bell Island, Washington 16
5. Lateral photo of the gammarid amphipod {Chromopleustes oculatus) used
for all laboratory and field experiments 26
6. Diagram ofhabitat positioning in amphipod choice experiments for aquaria
containing the dominant environments found at Bell Island: sand and shell
mixture. Laminarían algal blade, rock, and sea cucumber {Cucumaria
miniata) 33
7. Distribution ofmean body lengths for 36 images in testing pools containing
500 individuals 36
8. Comparison ofmean body lengths of Chromopleustes oculatus observed in
the field and those that were selected according to body size in the
laboratory experiments 39
9. Mean number of amphipods (abundance) observed for images taken in the
field at three locations (hase, center, and tip) at high and low slack tides on
the algal blade, Agarum fimbriatum 42
10. Mean nearest neighbor distances (expressed as body lengths) of
Chromopleustes oculatus at high and low slack tide on the algal blade,
Agarum fimbriatum 47
11. Mean nearest neighbor distances (expressed as body lengths) of
Chromopleustes oculatus at three locations (base, center, and tip) on the
algal blade, Agarum fimbriatum 49
12. Mean nearest neighbor distances (expressed as body lengths) of
Chrompleustes oculatus at three locations (base, center, and tip) at high and
low slack tides on the algal blade, Agarum fimbriatum 51
13. Mean nearest neighbor distances (expressed as body lengths) of
Chrompleustes oculatus at four densities (500, 1000, 2000, and 5000
individuals/0.664 m^) for testing pools in the laboratory to determine
impacts of density on spatial distributions 59
14. Mean nearest neighbor distances (expressed as mean body lengths) of
Chromopleustes oculatus at two time intervals (2 hr and 24 hr) at each of
the four densities for testing pools in the laboratory (500, 1000, 2000, and
5000 individuals/0.664 m^) 61
15. Mean nearest neighbor distances (expressed as mean body lengths) of
Chromopleustes oculatus in 0.664 m^ testing pools with and without
predators (excluding all distances < 1.37 body lengths, or 0.54 cm for
mating 66
16. Two-dimensional percentage of cover for each habitat in aquaria used for
habitat choice experiments; Cucumaria miniata, Agarum fimbriatum, rock,
sand and shell debris, and sand only 70
viii
INTRODUCTION
Many species use aggregate behavior for certain life history functions such as
reproduction, feeding, migration, and survival. Aggregations are defined as large
numbers of conspecifics in a particular area (e.g., swarms found on substrates or in the
water column). Determining distributions and patterns within an aggregated community
is crucial in understanding how the species maintains itself (Davis et al. 1991).
Crustaceans, especially amphipods of the Family Pleustidae, exhibit aggregate behavior,
which may be altered by abiotic and biotic factors such as temperature, salinity, life
cycles, water flow, suitable habitats, food, reproduction, in a few cases cannibalism,
predation, and inter- and intraspecific competition.
Abiotic Factors
Temperature regulates many aspects ofmarine species through habitat
preferences, migration, mating, and larval development (Maravelias 1997, Graham et al.
2001, Prince and Griffin 2001). Among gammarid amphipods, brood productivity and
time of development are directly related to temperature (Welton and Clark 1980,
Borowsky 1991). For example, female amphipods having a predicted longevity of 17-23
months, produce as many as six broods during a summer, with brood development time
decreased by more than half at increased temperatures (Mottram 1933, Hynes 1955,
Kinne 1960, Steele and Steele 1973, Welton and Clark 1980).
Changes in salinity do not appear to be a limiting factor in amphipod distributions
in their normal environments. For example, when two species of gammarid amphipods
2
are exposed to extreme salinity differences (above and below ambient), they show
survival times equal to experimental controls, suggesting that at least some amphipods
should be classified as euryhaline (Hopper 1960; also see Van Dolah 1978). However,
zooplankton (including amphipods) are positively affected by density discontinuities and
will aggregate at salinity fronts (Harder 1968). Fronts may act to concentrate food,
lending to aggregations, which may also impact mating (Straty 1972, Uye and Fleminger
1977, Facey and Van-Den-Avyle 1987, Wippelhauser and McCleave 1987, Parker 1995,
Graham et al. 2001, Grudemo and Andre 2001, Kingsford et al. 2002, Meade et al. 2002).
Biotic Factors
Biotic factors, in conjunction with abiotic factors, also can modify amphipod
behavior. For example. Van Dolah (1978) found that peak amphipod abundance is
correlated with both environmental and biological factors, with the former most
influential during the winter and the latter dominating other times of the year. Winter
declines in abundance at intertidal habitats correlate with amphipod subtidal migrations.
Additional causes for declines in intertidal populations include mortality due to freezing.
In contrast, mortality during the summer months coincides with limited food resources.
When food resources are limited, it is believed that adults displace juveniles from
preferred substrata. Mortality during the summer and fall also may be a result of
omnivorous fish predation on those amphipod species inhabiting palatable microhabitats
that are consumed by these fish.
3
Many gammarid amphipods swim upstream in currents (Macan and Mackereth
1957, Minckley 1964, Muller 1966, Kureck 1967, Lehman 1967, Hultin 1968, Elliot
1971). Gammarus bousfieldi is the only species (of six known to occur in the stream
system ofDoe Run, Kentucky) that appears to swim upstream in organized migrations
(Minckley 1964). A separate study on Gammarus pulex in the English Lake District
notes amphipods swim upstream in large quantities (Elliot 1971). Aggregates swim into
the current at this stony stream that ranges in turbidity from 7.1 cm*s'' in June to 95.0
cm*s'' in November (Elliot 1971).
Amphipods, like other species of crustaceans, physically adapt to areas of various
water flow. The literature suggests that amphipods found in extreme hydrodynamics
(e.g., swift currents) are excellent swimmers and specialize in their foraging mechanisms
(e.g., graze on algae itself or host organisms of the substratum) (Krapp 1993). Swimming
speeds typically show a linear relationship with body weight: larger amphipods swim
faster and vice versa (Takeuchi and Watanabe 1998). Swarms of small deep-sea
amphipods (< 5 cm compared with deep-sea amphipods much larger than 5 cm) are
found in the East Pacific Rise vent field from 2520 m to 2580 m (Kaartvedt et al. 1994).
Increased swimming speeds provide the individuals with the ability to maintain a position
in the current (Kaartvedt et al. 1994). The author of this study does mention that
“observations contrast with the general concept of low swimming activity in deep-sea
crustaceans ” (Kaartvedt et al. 1994).
Amphipods inhabiting slower moving waters with prevailing orbital motions cling
to substrate (Krapp 1993). These areas of oscillation zones are generally coupled with
4
hydroids and bryozoans; it is possible these omnivorous amphipods are commensal and
catch their prey out of the hydranths or polyps, and in some cases even biting off
tentacles of the host (Krapp 1993, David et al. 1996). Their ability to cling to the
substrate is excellent but they have limited swimming abilities with only short bursts of
speed for small distances (Krapp 1993).
For small herbivorous amphipods that live on algae or plants, food and habitat are
closely linked. Food choices in nature are constrained by the need to choose sites that
provide safe living. Many amphipods prefer to live on algae that are distasteful to
omnivorous fishes (Duffy and Hay 1991, Duffy and Hay 1994, Sotka et al. 1999, Poore et
al. 2000). The luxury of living and feeding on noxious plants allows amphipod species
such as Ampithoe longimana to reach their highest abundances during seasons when
predatory fish are most prevalent, while other amphipod species not living on noxious
plants decrease to near extinction (Duffy and Hay 1994). Therefore, there has been
selection for less mobile herbivores to be most tolerant of plant chemical defenses (Duffy
and Hay 1994).
Non-herbivorous amphipods living on chemically defended plants may also
benefit from the association. These commensal amphipods inhabiting macroalgae
typically do not forage on their defended hosts. Amphipods such as Ericthonius
brasiliensis wrap themselves in pieces of seaweed to create domiciles in which they
filter-feed (Sotka et al. 1999).
Aggregation facilitates reproduction. In organisms with external fertilization,
aggregation during gamete release increases fertilization success (Levitan and Petersen
5
1995, Coma and Lasker 1997, Levitan 1998, Styan 1998). Even for species with internal
fertilization, aggregation brings otherwise dispersed individuals together (e.g., Aguilar-
Perera and Aguilar-Davila 1996, Domier and Colin 1997, Sauer et al. 1997). These
aggregations may occur in areas in which no resources critical for reproductive success,
except mates, may be at stake; e.g., in a breeding system known as leks (e.g.,
Vehrencamp and Bradbury 1984, Anderson 1997).
Aggregation also affords the opportunity for cannibalism. There are numerous
reports of caimibalistic acts by marine animals including copepods (Anderson 1970).
Many species of freshwater Gammarus amphipods will become cannibalistic in
laboratory studies (Embody 1911, Sexton 1924, Sexton 1928, Clemens 1950, Jones 1951,
Schmitz 1967, Kostalos and Seymour 1976, Jenio 1980, Dick 1995, MacNeil et al. 1997).
Unfortunately, detailed feeding records for marine gammarid amphipods are rare due to
the complexity of the marine environment and size of the organisms (e.g., Muspratt 1951,
Larkin 1956, Macan 1965, Paine 1965, Nikolskii 1969, Manzi 1970). Some authors have
suggested that cannibalism erupts in individuals under severe stress, especially when
alternatives, such as dispersal, are not available (Colinvaux 1973, but see Fox 1975).
These stresses may include containment in laboratory systems (e.g., under severe
resource depletion or in situations where conspecifics are injured, diseased, and dead),
caimibalism ofjuveniles by adults, and cannibalism of smaller males by larger males
during mating season (MacNeil et al. 1999).
Gregarious behaviors serve many functional roles in defense against predators
including early warning, confusion, cooperative defense, and a dilution effect.
6
Individuals in an aggregation benefit from their own abilities to detect predators and also
from the aggregated sensory abilities of the group (Vulinec 1990, McCartt et al. 1997).
This early warning permits these organisms to take appropriate actions to reduce their
predation risk. For mobile organisms this may take the form of coordinated movements.
Predator confusion strategies are often observed within the masses by school circles,
splits, and flashes (e.g., Domenici and Batty 1994). In cooperative defense, aggressive
acts by groups drive off potential predators, a behavior that would be very risky if
performed by individuals (e.g., Vulinec 1990, Mueller 1994). Hamilton’s (1971) selfish
herd hypothesis proposed that individuals in the center of groups may suffer lower per
capita predation than solitary individuals or individuals on the periphery; a variety of
tests have provided some level of support for this theory (Treisman 1975; Turner and
Pitcher 1986; Sillen-Tullberg and Leimar 1988; Vulinec 1990, Krause 1993, but see
Parish 1989).
Aggregations may also benefit chemically defended prey and/or prey having
aposematic coloration. Mathews (1977) suggests that a predator’s consumption of large
numbers of a mildly noxious prey in rapid succession might build toxin levels to induce
an aversion not otherwise associated with that prey. In this case, aggregation reinforces
prior negative interactions between chemically defended organisms and predators.
Guilford (1985) updates an argument first developed by Fisher (1958) that aposematic
coloration and chemical defense are most likely to evolve in the presence of kin groups
(or under green beard selection).
7
Experimental tests demonstrate the value of aggregation in reducing predation
pressure on aposematic, chemically defended artificial prey (Alatalo and Mappes 1996).
In terrestrial insects in which it has been best studied, aggregation of conspecifics often
co-occurs with aposematic (poster) coloration and chemical defense (Sillen-Tullberg and
Leimar 1988). Larval aggregations are usually the result of clusters of eggs laid by single
females, thus creating kin groups (Sillen-Tullberg and Leimarl988, Vulinec 1990,
Bowers 1993). Gammarid amphipods, like other paracarid crustaceans, have direct
development; juveniles emerge from the marsupium without experiencing a planktonic
larval stage so common in other crustacean groups. Thus, for juvenile gammarid
amphipods at least kin groups may be present.
Recently, three species of chemically defended, aposematic, gammarid
amphipods {Chromopleustes oculatus, Chromopleustes lineatus, and Cryptodius kelleri)
have been discovered in the San Juan Islands, Washington (Figure 1) (Norton and
Stallings 1999). All species are sympatric at several locations. They occupy a variety of
•y
habitats and can reach densities greater than 24,500 individuals-m' (Norton and Stallings
1999, Aikins and Kikuchi 2001).
Fecundity in these amphipods may exceed 75 offspring per female (Hutchinson
2002), creating immediate kin groups. Amphipods, especially those in the family
Pleustidae, to which Chromopleustes belongs, also use color patterns for predator
defense, e.g., crypsis (Edmunds 1990, Stamp and Wilkens 1993), startle colorations
(Sargent 1990), and warning coloration (Bowers 1993). Pleustids also appear to use
coloration to mediate predator-prey interactions. For example, three species ofpleustid
8
Figure 1. San Juan Islands National Wildlife Refuge in Washington.
9
10
amphipods are known as Batesian mimics. These species mimic chemically defended
marine gastropods in coloration and locomotory behavior (Crane 1969, Field 1974,
Carter and Behrens 1980). Experiments with predatory fish demonstrate that mimicry
reduces attack rate (Field 1974). Similar mimicry of chemically defended organisms has
been described for other amphipod species (Goddard 1984). These marine amphipods
also exhibit behaviors comparable to many terrestrial insects; e.g., they share the limited
mobility characteristic of several caterpillar species.
A variety of field and laboratory data indicates that Chromopleustes oculatus,
Chromopleustes lineatus, and Cryptodius kelleri are chemically defended and
aposematic. Fishes that commonly prey upon other amphipod species avoid these three
species in the field. In laboratory experiments, fishes that capture these amphipods
typically release them unharmed (Hutchinson 2002). Predators appear to learn to
recognize and avoid these amphipods; experienced fishes will abort attacks on these
species (Zalewski 2001). The chemical defense that produces these effects appears to be
stored in the mid-gut caecae of these amphipods. This material is emitted when the
animal is stressed (e.g., when attacked, grabbed, or placed in a harsh environment). Niels
Lindquist (personal communication) at the Institute for Marine Studies (UNC-Chapel
Hill) tentatively identified the key compound as a terpene.
Chromopleustes oculatus, Chromopleustes lineatus, and Cryptodius kelleri are
found on available habitats by species-specific swarms. Tests to determine patterns of
distribution indicate at one spatial scale (comparisons of abundance patterns among 0.1-
m^ visual quadrats) all three species demonstrate high levels of aggregations, or clumping
11
(Norton and Stallings 1999). Tests using an Index ofDispersion - R (Donnelly 1978,
Krebs 1989) on C. oculatus at a finer spatial scale (within 0.0065-m^ photoquadrats)
indicate C. oculatus exhibits a more uniform or regular pattern on a multitude of habitats
found at Bell Island, Washington: rocks, sand-shell, dXïàAgarum fimbriatum. C. oculatus
maintains a minimum inter-individual spacing of 1 cm at high densities (60
individuals*65cm'^), a distance more than double the body length of an individual
amphipod (Figure 2). These earlier findings concur with Leising and Yen’s (1997)
studies performed on crustacean swarms. Leising and Yen (1997) note that several
species of invertebrate crustaceans will cohabitate in an area (swarm) but will remain
separated from conspecifics without any body-to-body contact.
This study concentrates on the spatial distributions of C. oculatus, the most
common amphipod found at Bell Island, Washington. At a fine spatial scale it is known
that large densities in excess of 60 individuals»65 cm‘^ of C. oculatus maintain a uniform
pattern of dispersion with inter-individual spacing of 1 cm among conspecifics on various
habitats at Bell Island (rocks, kelp, sea cucumbers, and sand-shell debris) (Norton and
Stallings 1999). No studies on the spatial distribution of C. oculatus have solely been
conducted on the dominant kelp habitat at Bell Island, Agarum fimbriatum, otherwise
known as the Colander Weed (Figures 3 and 4). Is A. fimbriatum the preferred substrate?
Or do C. oculatus aggregate on this kelp in response to distribution of food resources?
Additionally, the exact density that forces C. oculatus into this minimum inter-individual
spacing of 1 cm is not known.
12
Figure 2. Patterns of distribution for Chromopleustes oculatus from photoquadrats taken
summer 1999 at Bell Island on various microhabitats. Distributions are based on an
Index ofAggregation (R). n = 420 photoquadrats
13
dM(isctneaenimagchrebnsort)
Density (individuals / 65 cm^)
14
Figure 3. Detailed section taken from Figure 1 featuring Bell Island, located south of
Oreas Island and north of Shaw Island. Courtesy ofUSGS.
15
^^.vVicictim i KriMmj^
STATE PARK \
%í&lMd^Í
ihlhou
amd'f-
16
Figure 4. Image of the Colander Weed, Agarum fîmbriatum, the dominant spatial habitat
at Bell Island, Washington. Photographer unknown. Image courtesy of
WWW. life,bio.sunysb.edu/marinebio/kelp forest.html.
17
HYPOTHESES
The research described herein focuses on understanding aggregation in
Chromopleustes oculatus, the most common of the three defended species found in the
San Juan Islands. I examine inter-individual spatial distributions of amphipods in the
field using in situ underwater video. I focus on inter-individual distances of amphipods
on local kelp, investigating microhabitat of the blade at three locations (base, center, and
tip) at high and low slack tide. From the video, I look at density of amphipods in images,
in addition to amphipod spatial distributions using nearest neighbor distances.
Laboratory experiments further isolate the roles of density, the presence of predators, and
habitat availability on spatial distributions of C. oculatus.
The term distribution is defined here as inter-individual spacing among a group of
C. oculatus, calculated as nearest neighbor distances and further represented as the
number ofbody lengths. Density refers only to the number of individuals in a group
(swarm), on an image, or on a particular habitat. Patterns are the statistical nature of the
shape of the aggregate: clumped, uniform, or random. I asked the following specific
questions with regard to density and inter-individual distributions:
Distributions and Patterns
What is the in situ distribution of C. oculatus on its most abundant kelp habitat, A.
fimbriatum, at Bell Island? Is the distribution of amphipods affected by observation
times (high vs low slack tides)? Does distribution and/or patterns of the aggregate
19
change with changing density? Do differences exist in pattern and/or inter-individual
spacing in the presence ofpredators?
HI : There is no difference in mean amphipod nearest neighbor distance on A.
fimbriatum in the field as a function of amphipod location on the blade (base,
center, tip), tide state (high and low slack), or any interaction between the two
variables.
H2: There is no difference in mean nearest neighbor distance for densities
ranging from 753 individuals-m' to 7,530 individuals-m' .
H3: There is no difference in the mean nearest neighbor distance in the presence
of predators in the laboratory.
Density
At what density will C. oculatus maintain a regular spacing pattern of 1 cm in the
laboratory? Does density change within microhabitats on the blade ofA. fimbriatum
(base, center, and tip)? Given an array of habitats, would amphipods inhabit the same
preferred macroalgae in a controlled environment? Is habitat a possible reason for
aggregation?
H4: There is no difference in mean amphipod density on A. fimbriatum in the
field as a function of amphipod location on the blade (base, center, tip), tide state
(high and low slack), or any interaction between the two variables.
H5: The response of amphipods to habitat choices will be the same for each of
the four habitats: sea cucumbers, sand, rock, and sand-shell debris.
METHODS
Study Site
My base of operation was Friday Harbor Laboratories, a 484-acre facility on San
Juan Island administered by the University ofWashington. The primary study site was
Bell Island (48° 35' 47" N, 122° 58' 46" W) located in Wasp Passage, south of Oreas
Island and north of Shaw Island in the San Juan Islands, Washington (Figure 2). Bell
Island is characterized by mixed semidiurnal tidal cycles and swift currents due to
extreme changes in high and low tide (tidal differences of greater than 4 m in a 6-hour
period are common). Bell Island also is characterized by high levels of spatial
heterogeneity; abiotic conditions such as current, depth, and physical substrate vary at
fine spatial scales. As a result, the relative abundances of the dominant space-holding
organisms (e.g., laminarian algae, colonial tubeworms, sea anemones, and sea
cucumbers) also vary dramatically within microhabitats.
Chromopleustes oculatus, Chromopleustes lineatus, and Cryptodius kelleri have
patchy distributions at this site and can occur at relatively high densities (up to 24,500
individuals«m , personal observation) in subtidal habitats. In addition, they occupy a
variety of habitats including blades of laminarian algae, tentacles of sea cucumbers, other
benthic invertebrates, and shell debris (Norton and Stallings 1999).
Standardization
The spatial distributions of Chromopleustes oculatus analyzed in images taken at
various densities and in the presenee ofpredators in the laboratory were determined using
21
amphipods pre-selected for average body sizes of 0.39 cm to control for biological effects
with various cohorts. For all density experiments, a ruler was placed in the center of each
testing pool and photographed for scale after each image of the center was taken at 2 and
24 hours.
Nearest Neighbor Calculations
All images analyzed were captured in situ with a SONY DCR-PC 100 digital
video camera then downloaded from videotape using iMovie on a Macintosh G3. The
images were used to determine nearest neighbor distances using the T-square method
(Ludwig and Reynolds 1988, Krebs 1989). The T-square technique evaluated the
distribution of distances of amphipods from randomly selected points compared to
distances between each amphipod and its nearest amphipod neighbor. Specifically, the
coordinates of a point were chosen at random and the distance from the point to the
nearest amphipod was recorded. The distance from that amphipod to its nearest
amphipod neighbor was considered the nearest neighbor distance (NND). This process
was repeated for all amphipods for each of the randomly selected points.
Random coordinates (Xi,Yi) were generated in Microsoft Excel using a random
number formula and did not exceed the size of each image (length = 8.8 pixels, width =
6.6 pixels). A random number was produced from 0 to 8.8 pixels for Xi. An additional
random number was created between 0 and 6.6 pixels for Yj. For each image, a
maximum of 30 newly generated random numbers was created (Xi,Yi) and plotted on a
grid in Adobe Photoshop 5.5. From each random Xi,Yi coordinate I determined the
22
closest Chromopleustes oculatus and its coordinates (X2,Y2). From (X2,Y2) the nearest
neighbor coordinates (X3,Y3) were determined using the T-square technique. Then
nearest neighbor distances (pixels) were calculated using the equation:
Distance= V(x2-X3)"-(y2-y3)^
For each image, I used a maximum of 30 random numbers. In cases where I found fewer
than 30 amphipods on an image, the total of random numbers equaled the total number of
individuals visible («).
Determination of nearest neighbor distances was complicated by the fact that C.
oculatus appeared to mate (and possibly cannibalize) during my sampling season.
Therefore, all nearest neighbor distances less than 0.5 pixels were eliminated from the
calculations (Table 1).
Distances in pixels were converted to centimeters using images of a ruler taken in
situ for field observations and laboratory manipulations, respectively. For field
observations, a ruler was glued to an underwater slate to ensure all images were in focus.
The number of the blade was recorded on the slate below the ruler and was photographed
using the underwater video camera prior to photographing the base, center, and tip of the
preset algal blades. One cm equaled 0.889 pixels for field observations ofAgarum
fimbriatum. For lab experiments, a ruler was photographed in the center of the testing
pools once images had been taken of the five locations (north, south, east, west, and
center). One cm equaled 0.931 pixels for laboratory experiments.
23
Table 1. Percentage of nearest neighbor distances < 0.5
pixels (< 0.56 cm for field observations and < 0.22 cm
for laboratory manipulations) that were eliminated from
nearest neighbor distance calculations due to possible
mating or cannibalism for Chromopleustes oculatus.
FIELD OBSERVATIONS
AGARUM % Mating
Base 21.1
Center 17.3
Tip 13.2
Hi Slack 20.0
Low Slack 17.3
LAB MANIPULATIONS
DENSITY % Mating
500 11.9
1000 19.4
2000 42.9
5000 23.8
2hr vs 24 hr % Mating
500 2 hour 14.9
24 hour 6.5
1000 2 hour 14.6
24 hour 21.9
2000 2 hour 36.2
24 hour 57.5
5000 2 hour 23.8
24 hour 23.7
PREDATOR % Mating
with predator 7.5
without predator 11.9
24
Nearest neighbor distances (cm) were later expressed as body length
measurements (BL) for all graphs with respect to the field and laboratory results. I chose
body lengths because the anterior and posterior ends of amphipods were visibly clear in
all images, and average body sizes were relatively consistent for comparisons. Body
widths were not used because amphipods are laterally compressed, making width difficult
to measure.
Field Observations
A field survey was conducted on the mobility of Chromopleustes oculatus on
Agarum fimbriatum to determine if C. oculatus remained stationary during tidal cycles on
fixed blades. Blades were selected at random prior to beginning the field experiments
and counted for abundance from 13 June to 25 June 2000. A total of 12 blades were
selected along a 15-m transect at the northwest comer ofBell Island in water 8.0 m to
15.0 m deep. Divers on SCUBA counted whole algal blades for the number of C.
oculatus present on the dorsal and ventral side of the blade. Each blade was counted a
minimum of five times.
Field observations were made on the distribution of C. oculatus, the dominant amphipod
at Bell Island (Figure 5). Observations were made from 5 July to 23 July 2000 using
blades ofAgarum fimbriatum, a laminarían kelp, as habitat. Blades were selected at
random from a population on the northeast comer ofBell Island in water 11.7 m to 15.0
m deep along a 13-m transect. All observations were made using self-contained
underwater breathing apparatus (SCUBA). A total of 17 selected algal blades were
25
marked by attaching brightly colored flagging tape to the holdfast. Inter-individual
spacing was recorded in situ using a SONY DCR-PC 100 digital video camera in a Light
and Motion underwater housing. A small frame was attached to the camera housing to
ensure that images were standardized to a distance of 28 cm from the blade. Images were
taken during the day at high and low slack tides from three separate locations on each A.
fimbriatum blade: base, center, and tip. Each blade was observed at least five times over
the 18-day period, but it was not possible to census all 17 algal blades during each dive
due to time limitations for slack water intervals. Additionally, five blades were lost over
the course of observations.
Digital images were downloaded from the videotape and processed according to
the nearest neighbor calculations. The effects of tidal state and location on algal blades
on inter-individual amphipod nearest neighbor distances (+ SE) were evaluated by a
univariate analysis of variance (ANOVA). For all one-way ANOVAs, treatment means
were analyzed with an experiment-wise error rate of 0.05 using a data analysis feature in
Microsoft Excel 98. The interactions between tidal state and position on the blade were
studied using Two-way ANOVA. For two-way ANOVAs, Data Desk 6.0 was used for
both analysis and post-hoc comparisons.
26
Figure 5. Photo of Chromopleustes oculatus using scanning electron microscope (SEM).
Courtesy of Tim Charles.
27
28
Laboratory Experiments
The impacts of amphipod density, presence of predators, and habitat availability
on the distribution of Chromopleustes oculatus were tested in laboratory experiments.
Intact fronds ofAgarum fimbriatum containing large numbers of C. oculatus were
collected from the southern area ofBell Island using SCUBA. Clear plastic bags (70 cm
X 95 cm) were carefully placed over the tip and pulled to the base, where the holdfast was
removed from the rock and inserted into the bag. Collection bags were tied underwater
and taken to the surface to be transported to the laboratory for immediate separation. The
contents of each bag, including the seawater, were placed into a large tray (1 m^ x 0.3 m
deep) where C. oculatus individuals were caught using a turkey-basting bulb as a pipette.
Individuals were sorted by body sizes (mean body sizes used in the laboratory
experiments were approximately 0.39 cm).
All experiments were conducted outdoors in a 20-m^ arena containing static
seawater 0.25 m deep. A large plastic hood was constructed over the main area to block
direct sunlight and to prevent algal and diatom growth. A maximum of five round 0.664-
m X 23-cm deep plastic pools were placed in the arena to maintain ambient temperatures
of 9-12° C for each pool; the pools will be referred to as testing pools. Each testing pool
was filled with a homogenous sand substrate taken from a nearby hardware store. It was
not known if C. oculatus had preferential sites on A. fimbriatum; therefore, this sand
substrate was used as a control in the testing pools to test for changes in nearest neighbor
distances with respect to density and the presence of predators without complications of
habitat choices. Additionally, A. fimbriatum was not used in experiments because of the
29
rapid decomposition of the algal blade when removed from the study site and its possible
negative affects on C. oculatus in static waters. Between trials (after 24 hours), both the
sand and testing pools were rinsed with freshwater and the exposed pool edges were
scrubbed with a nylon brush. In addition to cleaning the testing pools between trials, C.
oculatus were moved to a single holding pool and fed brine shrimp pellets for a period of
24 hours to again be retested in subsequent trials.
Spatial distributions and patterns of aggregates. Amphipod densities were
manipulated to determine changes in nearest neighbor spacing. The testing pools with
homogeneous sand substrate were filled with seawater in a static system. Four density
treatments were chosen to depict a higher and lower range of densities most commonly
observed in the field on various habitats: 500, 1000, 2000, and 5000 individuals per pool,
for densities of 753 individuals-m' to 7,530 individuals-m' . Testing pools were not
saturated with amphipods. In order to completely fill a testing pool to saturation,
approximately 43,681 amphipods would need to be added. To fill the testing pool to the
highest density recorded at Bell Island, approximately 16,268 individuals would need to
be added to the testing pool. Numbers of this magnitude were not replicated in testing
pools due to the ecological impacts of removing large quantities of the population at Bell
Island, since amphipods were not released after each density experiment. Amphipods
remained in holding pools and were fed brine shrimp to be used in subsequent
experiments. The maximum number of amphipods kept in a holding pool at any time
was approximately 6,000 individuals. Individuals were removed from the holding pools
using a turkey-basting bulb and the appropriate numbers of individuals were placed into
30
testing pools. Trials in the testing pools lasted 24 hours. The testing pools remained in
the larger holding arena to maintain ambient temperatures of 9-12° C. Replicates did not
all occur on the same day; the four density treatments were conducted during a period of
several weeks (14 June to 11 July 2000) and were conducted at random times during the
day, in some cases to allow for time to dive at the field site.
Images were recorded on digital videotape at 2-hour and 24-hour intervals after
initial introductions of the amphipods to each pool. At each time interval, five images
were taken at non-overlapping locations in the pools: the four compass directions (north,
south, east, west) and the center. There was a minimum of four trials per density
treatment. A trial was defined as two images taken within a 24-hour period (n = 21 total
trials: 5 trials at 500, 8 trials at 1000, 4 trials at 2000, and 4 trials at 5000).
Nearest neighbor distances were calculated using the method outlined previously.
Patterns in the arrangement of amphipods in testing pools were assessed using the
plotless technique outlined in Holgate (1965) and Brower et al. 1998. A single univariate
ANOVA was used to test changes in average nearest neighbor distances (+ SE) in
density. An additional univariate ANOVA examined the density distributions over time.
Results from the ANOVAs were compared using an error rate of 0.05.
Predators and spatial distributions. The testing pools with sand substrate were
again placed in the larger holding arena to maintain ambient temperatures. This time,
testing pools contained 500 amphipods, taken from holding pools, and one Oligocottus
maculosus, a naïve predatory fish captured from a nearby tide pool. O. maculosus is a
predatory sculpin that frequently feeds on palatable amphipods and had not been
31
previously exposed to the chemically defended amphipod species, Chrompleustes
oculatus. Prior to adding the predator, amphipods were given 15 minutes in each pool to
establish equilibrium. After 15 minutes, the single O. maculosus was introduced and
inter-individual distributions were again recorded on digital videotape. Recording
continued thereafter every 15 minutes for a total of six images per trial. At each
recording, five images were taken at the four compass directions and center in non-
overlapping locations. The experiment was repeated 13 times (n=390 images: 13 trials x
5 locations x 6 images per trial). The density treatment experiment using 500 individuals
per testing pool was considered the control in the absence of predators. The methods for
the control are outlined previously for testing pools in Spatial Distributions and Patterns
ofAggregates.
Nearest neighbor distances of amphipods in the presence of predators were
calculated using the method outlined previously. Then, nearest neighbor distances were
compared with the images taken from the testing pools containing 500 individuals-0.664
m'^ (n = 50 images) and the images ofpools containing the predator (n = 390 images). A
single univariate ANOVA was used to test differences among average nearest neighbor
distances (+ SE) in the presence or absence of a predator. Results from the ANOVA
were analyzed using an error rate of 0.05.
Habitat choice experiments. I examined habitat choice by introducing
Chromopleustes oculatus into two different sized aquaria (Aquarium A: 0.119 m^ x 25
cm deep. Aquarium B: 0.139 m^ x 28 cm deep), each contained the dominant field
habitats: sea cucumber (Cucumaria miniata), Agarum fimbriatum, sand and shell
32
mixture, and rock (Figure 6). Habitats were collected from Bell Island prior to the
beginning of the experiment and placed in flow-through sea water tables at ambient
temperatures of 9 - 12° C. Before initiating each test the aquaria were cleaned with a
nylon brush, and the sand and sand-shell mixtures were rinsed with seawater and placed
so that each covered 50% of the bottom. I then added to each aquarium a fresh piece of
algae, a fresh-collected C. miniata, and a rock rinsed with seawater.
Approximately 50% of the bottom of each aquarium was covered with sand (used
in the previous laboratory experiments) and the other 50% was a natural sand and shell
mixture from Bell Island. The rock and the piece of algae were then placed in the center
of the tank while the C. miniata was positioned in a comer. Areas for each habitat were
measured two-dimensionally in NIH-Image 1.63. All aquaria were set up with once-
through flows of seawater. Habitat choices of 50 amphipods were monitored at 2, 6, and
24 hours after introduction. The experiment was repeated four times (n = 21 images: 4
trials X 3 time intervals x 2 aquaria). Three images were lost due error (image was too
dark or was not taken). The expected amphipod values were calculated by multiplying
the percentage of the total area occupied by each of the five habitats with the 50
amphipods placed in each aquarium (number of expected values, n = 105; 21 images x 5
habitats). Choices were compared for the number of amphipods found on each habitat
using a Chi-Square analysis in Microsoft Excel 98.
33
Figure 6. Diagram depicting an example of habitat positioning used in Aquarium A and
B in habitat choice experiments.
34
RESULTS
Standardization
Ninety-two percent of images in testing pools containing 500 individuals had
body lengths within 0.1 pixels of the average body size used for laboratory experiments.
Average body length per image (based on 173 individuals) ranged from 0.160 pixels to
1.566 pixels (Figure 7). The mean body length for the 36 images was 0.313 + 0.04 pixels
(+ values are standard error). The average body length of individuals selected for testing
pool experiments was 0.39 cm, which is equivalent to body lengths of 0.363 pixels (1 cm
= 0.931 pixels for laboratory calculations).
Field Observations
Results from the field survey, conducted prior to the field observations, suggested
Chromopleustes oculatus were highly mobile on individual blades. The mean number of
amphipods on any blade ranged from approximately 14 + 2.7 individuals to 358 + 23.4
individuals (Table 2).
The average body length for adults and juveniles in the field, determined from six
individuals in three random pictures, was approximately 0.46 + 0.03 cm (Figure 8).
Populations of similar sizes of Chromopleustes oculatus (referred to hereafter as cohorts)
varied in body sizes throughout the summer at Bell Island. Late May cohorts were
36
Figure 7. Distribution ofmean body lengths for 36 images in testing pools containing
500 individuals. Distribution is shown according to number of observations for each
body length.
37
38
Table 2. Amphipod variability, expressed as Mean + SE,
from 12 algal blades. Counts were taken as a field survey
in situ at Bell Island 13 June 2000 through 25 June 2000
during high and low slack tide prior to beginning field
observations. Blade numbers represent the assigned
number given to each blade in the field, n = # of
observations per blade
Blade n Mean + SE
12 1 358.4 ±23.4
11 5 215.8 ±22.4
9 7 292.4 ± 17.4
s5 5 38.0 ± 10.8
s4 5 115.4 ±.6.1
7 5 169.6^19.0
6 7 13.9 + 2.7
10 6 282.7 4^18.1
s2 5 83.4±7.0
si 5 34.2^3.5
5 6 86.0 4^5.9
8 8 220.8 + 28.6
39
Figure 8. Comparison ofmean body lengths of Chromopleustes oculatus observed in the
field and those that were selected according to body size in the laboratory experiments (n
= 18 individuals; 6 individuals per image x 3 images).
40
Field Laboratory
Location
41
approximately twice as large as cohorts selected in mid-June. By mid-June average body
sizes seen in the field were proportional to those used in the laboratory observations. All
individuals were clearly visible in the pictures; absolute underwater visibility ranged
from limited (~ 1 m) to moderate (~ 4 m) at slack tide. Best visibility occurred at high
slack tide. Amphipods remained in place at slack tide unless touched by another
organism (e.g., amphipod, cohabitant fish, etc.) or a diver disturbed the blade.
The mean abundance of amphipods found on the base of the 17 Agarum
fimbriatum blades was 20 individuals + 2.91 SE (standard error) at high slack tide and 20
+ 4.07 individuals at low slack tide (Figure 9). This mean abundance was equal to a
mean density of 3,076 amphipods per m^ for the base of the blade. Mean abundance at
the center of the blade was 13 + 1.77 individuals at high slack tide and 12 + 1.63
individuals at low slack tide, for a mean density of 1,923 amphipods per m . Mean
amphipod abundance at the tip of the blade was 8 + 2.23 at high slack tide and 7 + 1.86 at
low slack tide, for a mean density of 1,153 amphipods per m .
Amphipod abundance was influenced by position on the blade but not by tidal
state. A 2-way ANOVA showed abundances did not vary as a function of tide state
(Table 3,p = 0.7; Figure 9). The interaction between position and tide state also was
non-significant (Table 3,p = 1, Figure 9). Further analysis using a Bonferroni Post Hoc
42
Figure 9. Mean number (+ SE) of Chromopleusîes oculatus for images taken in the field
on Agarum fimbriatum at three locations — hase, center, and tip — at high and low slack
tides, n = number of images.
43
Base Center Tip
High Low High Low High Low
44
Table 3. ANOVA of the differences in mean number of individuals at three positions on
Agarum fimbriatum using images taken at high and low slack tides of natural populations
of Chromopleustes oculatus at Bell Island, n = number of images.
Source df n MS F P
Blade Position (Base / Center / Tip) 2 161 2117.8000 8.99 0.0002
Tide (High Slack / Low Slack) 1 161 34.0864 0.14 0.7041
Interaction (Tide / Blade Position) 2 161 1.2507 0.01 0.9947
45
test indicated densities observed at the base were significantly different from the center (p
= 0.01) and tip {p = 0.0002; Table 4). The center and tip did not show any statistical
differences (p = 0.3).
Mean nearest neighbor distances (MNND) in graphs were expressed as distance
in body lengths, calculated for the average body length seen in the field (0.46 cm = 1
body length). The mean nearest neighbor distance between amphipods on kelp blades
ranged between 3.24 and 4.63 body lengths (BL). Excluding blade positions, the MNND
at high slack tide and low slack tide was 3.56 + 0.08 BL and 3.71 + 0.08 BL, respectively
(Figure 10). Excluding tidal state, the MNND at the base was 3.41 + 0.07 BL, at the
center was 3.66 + 0.09 BL, and the tip was 4.46 + 0.18 BL (Figure 11). The MNND
found on the base of the A. fimbriatum blade was 3.24 + 0.09 BL at high slack tide and
3.54 0.12 6L 3.t lo^v slâ.ck tide (3.II ^ V3.1iies sre st^tid3.rcl errors^ Figure 12), The
MNND at the center of the blade at high slack tide was 3.73 ± 0.14 BL and MNND at
low slack tide was 3.59 + 0.11 BL. The MNND at the tip of the blade was 4.29 + 0.21
BL at high slack tide and 4.63 + 0.29 BL at low slack tide.
A 2-way ANOVA showed MNND did not vary as a function of tidal state (Table
5,p = 0.7; Figure 10), but MNND shifted as a function ofposition on the blade (Table 5,
p < 0.0001 ; Figure 11). The interaction between blade position and tide state also was
46
Table 4. Bonferroni Post Hoc test comparing the mean abundances of Chromopleustes
oculatus at three locations on Agarum fimbriatum. n = number of images.
Comparison n Difference Std. Error P
Base - Center 121 -8.2954 2.845 0.0122
Base - Tip 97 -13.1961 3.222 0.0002
Center - Tip 104 -4.9007 3.124 0.3157
47
Figure 10. Nearest neighbor distances (NND) expressed as mean body lengths (+ SE) of
Chromopleustes oculatus on Agarum fimbriatum at high and low slack tides. Body
lengths <1.22 (0.56 centimeters) were removed from the calculations, n = number of
NND.
48
0.0
High Slack Tide Low Siack Tide
49
Figure 11. Nearest neighbor distances (NND) expressed as mean body lengths (+ SE) of
Chromopleustes oculatus on Agarum fimbriatum at three locations — base, center, and tip
— independent of tidal cycle. Body lengths <1.22 (0.56 centimeters) were removed from
the calculations, n = number ofNND.
50
(NLBeNnogDtdh)y
Base Center Tip
51
Figure 12. Nearest neighbor distances expressed as mean body lengths (+ SE) of
Chromopleustes oculatus on Agarum fimbriatum at three locations — base, center, and tip
— at high and low slack tides. Body lengths <1.22 (0.56 centimeters) were removed
from the calculations, n = number ofNND.
52
Base Center Tip
High Low High Low High Low
53
significant (Table 5,p = 0.0009; Figure 12). In further analysis using a Bonferroni Post
Hoc, I found the following specific interactions to be significant; interactions between the
base and tip of the blade during tidal state (high and low, and high versus low,
respectively), interactions between tidal state at the given position, center (p = 0.04), and
interactions between the position, center and tip, of the Agarum blade at high and low
slack tides, in addition to low slack tide (Table 6). There was no interaction between the
center and tip at high slack tide {p = 0.3). Additional Bonferroni comparisons
specifically for blade position showed the tip was statistically different from the center
(Table 6, p < 0.001) and also from the base (Table 6,p< 0.001), but the base and center
were not statistically different (p = 0.9).
Laboratory Experiments
The MNND values for the laboratory experiments were much lower than the
normal avoidance minimum of 1 cm established from previous experiments (Norton and
Stallings 1999) because ofmating and cannibalism attempts. Amphipods appeared to be
attempting to mate in testing pools. One amphipod, presumably the male, would position
itselfperpendicular to the female, “grabbing” her with his claws. The male would
palpate the female by brushing its pleopods rapidly against the female’s body. Additional
observations relevant to the following laboratory results were that approximately 10-15
54
Table 5. ANOVA of the differences in mean nearest neighbor distances at three positions
on Agarum fimbriatum using images taken at high and low slack tides of natural
populations of Chromopleustes oculatus at Bell Island, n = number of nearest neighbor
distances.
Source df n MS F P
Blade Position (Base / Center / Tip) 2 1763 19.9452 22.08 <0.0001
Tide (High Slack / Low Slack) 1 1763 0.1480 0.16 0.6858
Interaction (Tide / Blade Position) 2 1763 6.3726 7.05 0.0009
55
Table 6. Bonferroni Post Hoc interaction results for mean nearest neighbor distances that
shifted as a function of tidal state (High / Low Slack tide) at a given blade position (Base
/ Center / Tip) for Chromopleustes oculatus.
Comparison 1 Comparison2 Difference Std.Error P
High, Center High, Base 0.2011 0.0712 0.069545
High, Tip High, Base 0.4270 0.0918 0.000053 ***
High, Tip High, Center 0.2259 0.0978 0.272335
Low, Base High, Base 0.1215 0.0643 0.598938
Low, Base High, Center -0.0797 0.0726 0.991537
Low, Base High, Tip -0.3056 0.0928 0.015135 *
Low, Center High, Base -0.0164 0.0650 1.00000
Low, Center High, Center -0.2175 0.0732 0.044278 *
Low, Center High, Tip -0.4434 0.0933 0.000033 ***
Low, Center Low, Base -0.1378 0.0665 0.444284
Low, Tip High, Base 0.5861 0.1013 0.000000 ***
Low, Tip High, Center 0.3850 0.1067 0.004768 *
Low, Tip High, Tip 0.1591 0.1214 0.957833
Low, Tip Low, Base 0.4647 0.1023 0.000088 ***
Low, Tip Low, Center 0.6025 0.1027 0.000000 ***
56
amphipods in each testing pool would die before the 24 hours of testing ended. In testing
pools only, I noted several cases of cannibalism where amphipods would attack a single
dead or dying individual. At later observation times only the exoskeleton would remain
of the injured individual. Therefore, the NNDs < 0.5 pixels (<1.22 body lengths (0.56
cm) for field observations and <1.37 body lengths (0.54 cm) for laboratory observations)
were not used in the comparisons ofmeasured MNND (Table 1). Any distance less than
the minimum 0.5 pixels was regarded as reproductive efforts or cannibalism and
consequently was removed from calculations. Additionally it was noted that during July,
small juveniles (-0.2 cm) could be found in testing pools with adults. These pools had
not been replenished with amphipods from the field for several weeks.
Spatial distributions and patterns of aggregates. The average body length
amphipods used in the lab, determined from six individuals in three random pictures, was
approximately 0.39 + 0.03 cm (Figure 8). Mean nearest neighbor distances (MNND) in
graphs were expressed as distance in body lengths (BL), calculated for the average body
length used in the lab (0.39 cm = 1 body length). The mean nearest neighbor distances
(MNND) were inversely related to density of amphipods in the testing pools. The
MNND in testing pools containing 500 individuals (for a density of 753 individuals* m'^)
was 5.98 + 0.2 BL (all + values are standard errors), and ranged from 6.42 + 0.3 BL at
two hours after introduction to 5.25 + 0.3 BL 24 hours after introduction. The MNND in
testing pools containing 1,000 individuals (for a density of 1,506 individuals* m' ) was
4.7 ±0.1 BL, and ranged from 5.34 + 0.3 BL to 4.36 ±0.1 BL two and 24 hours after
introduction, respectively. TheMNND in pools containing 2,000 individuals (for a
57
9 9
density of 3,012 individuals* m' ) was 3.7 + 0.1 BL or 3, 012 individuals per m , and
ranged from 3.65 + 0.1 BL to 3.84 + 0.2 BL at two and 24 hours, respectively. The
MNND in testing pools containing 5,000 individuals (for a density of 7,530 individuals*
m'^) was 4.70 ±0.1 BL or 7,530 individuals per m^, and ranged from 4.50 ± 0.2 BL to
4.92 + 0.2 BL at two and 24 hours, respectively.
These differences in mean nearest neighbor distance as a function of density were highly
significant (Table 7, p < 0.0001; Figure 13). Comparisons ofMNND at two and 24 hours
after introduction were also significant (Table 7, p = 0.0237; Figure 14). The Bonferroni
Post Hoc showed all but one test for the MNND were significantly different for the four
densities; the comparison made between the MNND in testing pools containing 1,000 and
5,000 individuals was not significant (Table 8, p = 0.947). Bonferroni Post Hoc results
demonstrated all but one comparison (contrasts between MNND at 2,000 and 5,000
individuals-0.664 m'^) showed statistical differences when MNND were compared using
2-hour observation times as a constant for the four densities (Table 9). There were no
statistical differences in MNND comparisons when the observation time, 24 hours, was
used as constant. In comparisons where no constant was used, 2 hours versus 24 hours
and vice versa, significant differences were only found forMNND at 500
individuals O.664 m‘^ and2,000 individuals-0.664 (Table 9).
58
Table 7. ANOVA of the relative change in mean nearest neighbor distances of
Chromopleustes oculatus at 2 and 24 hours in experiments manipulating four densities
(individuals per 0.664m^).
Source df n MS F P
Time (2 vs 24) 1 2003 4.15977 5.1250 0.0237
Density (500, 1000, 2000, 5000) 3 2003 15.0444 18.535 <0.0001
Interaction (Time / Density) 3 2003 7.74631 9.5437 <0.0001
Table 8. Bonferroni Post Hoc comparisons ofmean nearest neighbor distances analyzed
from images of Chromopleustes oculatus taken at five preset locations in testing pools for
four densities, respectively (500, 1000, 2000, and 5000 individuals 0.664m^).
Comparison Difference Std.Error P
1000 - 500 -0.295241 0.0661 0.000050 ***
2000 - 500 -0.544271 0.0735 0.000000 ***
2000 - 1000 -0.249031 0.0575 0.000094 ***
5000 - 500 -0.337438 0.0669 0.000003 ***
5000 - 1000 -0.042197 0.0488 0.947290
5000 - 2000 0.206833 0.0585 0.002501 *
59
Figure 13. Nearest neighbor distances (NND) expressed as mean body lengths (+ SE) of
Chromopleustes oculatus at four densities (individuals/0.664m ). Body lengths <1.37
(0.54 centimeters) were removed from the calculations. n= number ofNND.
7
500 1000 2000 5000
Density (individuals/0.664m2)
61
Figure 14. Nearest neighbor distances (NND) expressed as mean body lengths (+ SE) of
Chromopleustes oculatus at 2 time intervals—2 hours (unshaded) and 24 hours (shaded)—
for 4 densities. Body lengths <1.37 (0.54 centimeters) were removed from the
calculations. n= number ofNND.
62
(NLBeNnogDtdh)y
500 1000 2000 5000
Density (individuals/0.664m2)
63
The patterns of the aggregates in testing pools were significantly different than
random (p = 0.05), with the exception of 2000 individuals-0.664 m'^. The plotless index
of aggregation by Holgate (1965) showed patterns of aggregates in testing pools
containing 500, 1000, and 5000 individuals-0.664 m'^ were contagious, or clumped.
Predators and spatial distributions. Chrompleustes oculatus were not visibly
disturbed by Oligocottus maculosus unless touched. In many cases C. oculatus would
attach to the fish’s tail, and occasionally to its body. This caused the fish to swim
violently in efforts to remove the amphipod(s).
Mean nearest neighbor distances (MNND) in graphs were expressed as distance
in body lengths (BL), calculated for the average body length used in the lab (0.39 cm = 1
body length). The MNND for testing pools that contained a single O. maculosus was
6.48 ±0.1 BL (all ± values are standard errors; Figure 15). The MNND for testing pools
that did not contain a predator was used from the density experiment (MNND of 500
individuals = 5.98 BL ± 0.2). A univariate ANOVA found the differences in MNND in
the presence or absence ofpredators was significant (p = 0.05).
64
Table 9. Bonferroni Post Hoc interaction results for mean nearest neighbor distances that
shifted as a function of density at a given time for Chromopleustes oculatus (comparisons
are sorted by increasing probabilities).
Comparison 1 Comparison2 Difference Std.Error P
1000-24 500-2 -0.617366 0.0809 0.000000
2000 - 2 500-2 -0.779830 0.0875 0
2000 - 2 1000-2 -0.454916 0.0752 0.000000
2000 - 24 500-2 -0.660510 0.1040 0.000000
5000 - 2 500-2 -0.575988 0.0860 0.000000
5000 - 24 500-2 -0.450685 0.0859 0.000005
5000 - 24 2000 - 2 0.329144 0.0726 0.000170
1000-24 1000-2 -0.292452 0.0674 0.000415
2000 - 2 500 - 24 -0.428033 0.1040 0.001113
1000-2 500-2 -0.324913 0.0882 0.006567
2000 - 24 1000-2 -0.335597 0.0939 0.009985
5000 - 2 1000-2 -0.251075 0.0734 0.017796
500 - 24 500-2 -0.351797 0.1137 0.054476
5000 - 2 2000 - 2 0.203841 0.0726 0.132266
1000-24 500 - 24 -0.265568 0.0984 0.179089
2000 - 24 500 - 24 -0.308713 0.1182 0.224631
5000 - 24 1000-24 0.166680 0.0644 0.239112
2000 - 2 1000-24 -0.162464 0.0665 0.337893
5000 - 24 2000 - 24 0.209825 0.0918 0.468647
65
Table 9. Continued
Comparison 1 Comparison2 Difference Std.Error P
5000 - 2 500 - 24 -0.224191 0.1027 0.562627
5000 - 24 5000 - 2 0.125303 0.0707 0.892638
5000 - 24 1000-2 -0.125772 0.0734 0.921122
2000 - 24 2000 - 2 0.119319 0.0932 0.998120
5000 - 24 500 - 24 -0.098888 0.1026 0.999989
5000 - 2 2000 - 24 0.084522 0.0918 0.999996
1000-2 500 - 24 0.026884 0.1045 1
2000 - 24 1000-24 -0.043145 0.0870 1.00000
5000 - 2 1000-24 0.041377 0.0645 1.00000
66
Figure 15. Nearest neighbor distances (NND) expressed as mean body lengths (+ SE)
among 500 Chromopleustes oculatus in 0.664m^ testing pools with and without predators
(excluding all distances < 1.37 body lengths, or 0.54 centimeters for mating). n= number
ofNND.
67
(LNBeNnogDtdh)y
With Predator Without Predator
68
Habitat choice experiments. Three images were lost due to error: one image
was overexposed and one set of images at 24 hours was not taken of the two aquaria.
There were a few cases where all amphipods were not accounted for in images (total ^
50). This error resulted from individuals attaching to the glue holding the aquarium
together. The seams of the aquarium were not accounted for in the two-dimensional
technique to assess the percent of space each habitat occupied. In one image taken at 24
hours in Aquarim B, C. miniata was coiled into a comer and was not feeding. In this
image, no amphipods were found on any of the provided substrates, but were attached to
the glue of the aquarium.
When presented a choice, the amphipods preferred cucumber habitat to kelp
blades, sand, rock, or sand and shell mixtures. The average percentage of habitat that
Cucumaria miniata occupied was 6.8 + 2.4% (all + values are standard deviations) and
the average abundance of amphipods found on C. miniata was 18 + 9 individuals. A.
fimbriatum represented 20.3 + 4.0% of the available habitat, but the average abundance
of amphipods found on the blade was only 2 + 2 individuals. Rock was 9.5 + 2.4% of the
habitat, and average abundance of amphipods found on rock was only 1+2 individuals.
Sand and shell mixture represented 28.1 + 7.7% of the habitat and sand represented 35.4
+ 6.4%, yet average abundance on the sand and shell was only 5 + 5 individuals and 0+1
individual on the sand. A Chi-Square test showed the amphipod population (n=50) was
not normally distributed on habitats (X^ = 2543, df= 20, p <0.005) (Table 10; Figure 16).
69
Table 10. Table of values used in Chi-Square analyses: two-dimensional percentage of
cover for each microhabitat in aquaria, expected number of Chromopleustes oculatus
with respect to habitat percentages, and observed numbers of individuals in aquaria.
Expected and Observed are the average number of individuals for two aquaria taken from
introductions of 50 amphipods. (n = 21 images)
Microhabitat % Coverage Expected Observed
C. Miniata 7 3 18
Rock 9 5 1
Sand-Shell 28 14 5
A. fimbriatum 20 10 2
Sand only 35 18 0
70
Figure 16. Two-dimensional percentage of cover (+ standard deviation) for each habitat,
n = 21, number of replicates.
71
50
45
40
Cucumaría miniata Agarum fímbriata Rock Sand and Shell Sand only
Microhabitat Choices
DISCUSSION
Standardization
I concluded that no standard was required for processing images taken of
amphipods in testing pools. Results indicated that inconsistent camera distances had little
variability in average body lengths for images (0.313 pixels) compared with average
body lengths selected for laboratory experiments (0.363 pixels; Figure 2). I assumed
standardization would not be necessary for other testing pools that invariably had smaller
errors in MNND (Figure 10).
Field Observations
In designing the original field experiments, I had assumed that Chromopleustes.
oculatus remained stationary during much of the tidal cycle. In an earlier research
design, based on the assumption of little to no mobility, I proposed to introduce and
deplete amphipods from existing swarms in situ to manipulate densities and monitor
nearest neighbor distances in natural populations during a tidal cycle. Given the mobility
of C. oculatus, the field work was limited to observations at low slack and high slack tide
for individual A.fimbriatum blades (Table 2). The morphological constraints of the
amphipods (apparent limited swimming capabilities and laterally compressed bodies) and
swift currents at Wasp Passage likely contributed to prior observations that amphipods
were immobile and best suited to reside in more complex habitats offering greater
opportunity for good attachment and shelter (Pavia and Aberg 1998). Each blade of
Agarum fimbriatum has many indentations and the effects of currents on the algal blade
73
create a whipping motion at the tip of the blade, much like a flag flapping on a windy
day. The blades stream easily in the swift current, but remain motionless at slack water.
Previous research ofwater flow effects on Laminaria in high-energy systems suggest
flow is mostly turbulent around larger macroalgae, and turbulent boundary layers are
formed more rapidly among single-bladed kelps (Hurd and Stevens 1997). The
swimming speed is not known for C. oculatus, nor is the water velocity at which these
amphipods can withstand simply by clinging to available substrates. Studies condueted
on deep-sea gammarid species have observed smaller amphipods (2-3 cm in length)
swimming vigorously at 5-10 cm»s'' into the current, only to maintain their position in
the water column without forward advancement (Kaartvedt et al. 1994). Takeuchi and
Watanabe (1998) also report that swimming speeds have a linear relationship with body
weight, suggesting the swimming speed ofC. oculatus (body lengths of ~ 0.4 cm) would
be much less than 5-10 cm»s''.
My held survey on whole blade counts at slack tide suggested amphipods were
mobile, either directly by activity or indirectly via currents, given abundance changes
between successive observations. Observations of amphipods on whole blades ofA.
fimbriatum taken at various slack tide intervals had shown the number of individuals
varied in abundance by as much as 81% on the marked algal blades. This study did not
focus on times at which C. oculatus were active (slack tide versus tidal exchanges);
therefore, it is not known if the amphipods were moving during tidal exchanges. There
was no statistical analysis performed on the significance of the amphipod mobility (e.g.,
tide to tide comparisons) since it was only necessary to determine if individuals would
74
remain on blades for one tidal cycle to observe changes in their spatial distributions
(Table 2).
Spatial distributions. HI : There is no difference in mean amphipod nearest
neighbor distance on Agarum fimbriatum in the field as a function of amphipod location
on the blade (base, center, and tip). Differences in MNND were found to occur among
amphipod positions on A. fimbriatum. The MNND at the tip of blade was higher than the
MNND at the center and base. The MNND at the base approached a minimum
separation distance of 1 cm, a distance more than double an individual’s body (body
lengths of 0.46 cm observed in the field). These results are similar to other experiments
conducted on several species of copepods that maintain minimum separation distances
(Leising and Yen 1997).
HI : There is no difference in mean amphipod nearest neighbor distance on A.
fimbriatum in the field as a function of amphipod location at tidal state (high and low
slack tide). I failed to reject this null hypothesis. There were no shifts in distribution or
abundance in images as a function of tidal state (high and low slack tide) with respect to
images taken in the field. The absence of shifts in MNND also suggests that tidal state
does not reinforce chemical defenses for C. oculatus.
Density. H4: There is no difference in mean amphipod density on Agarum
fimbriatum in the field as a function of amphipod location on the blade (base, center, tip),
tide state (high and low slack), or any interaction between the two variables. The
changes in distances noted above were coupled with changes in density. As density
increased, MNND decreased. Amphipods were concentrated at the base of the blade and
75
just as previous research on Chromopleustes oculatus has indicated (Norton and Stallings
1999), I found the group’s members decreased inter-individual spacing in large
populations, but still maintained a minimum separation distance of 1 cm. The base is
nearest the holdfast and provides the least movement during tidal exchanges. Mean
abundance decreased towards the tip, typically the most actively moving portion of the
blade. The reduced abundances seen at the tip may also be influenced by slack tide.
Slack tide at the site lasts only for approximately 45 minutes, which may not be enough
time to enable amphipods to reposition themselves at the tip, despite rapid colonization
(Pavia et. al. 1998). Leising and Yen (1997) reported similar results from copepod
experiments testing swarms of individuals. As the density of the copepod swarms
increased, the average nearest neighbor distance decreased (Leising and Yen 1997).
One could argue that MNND seen at the base of the blade supports Hamilton’s
(1971) selfish herd theory. Individuals reduced nearest neighbor distances in the
presence of conspeciflcs in attempts to decrease the probability that any single individual
would be attacked. Tighter spacing within the aggregate may also serve to increase the
amount of chemical defenses found within a small area. Increased concentrations of C.
oculatus at the base of the blade supports Mathews (1977) argument on aggregations of
noxious individuals. C. oculatus are the least of the three defended species of amphipods
found at Bell Island, Washington (Hutchinson 2002) and a predator’s consumption of
large numbers of a mildly noxious prey in rapid succession might build toxin levels to
induce an aversion not otherwise associated with that prey, reinforcing negative
interactions between chemically defended organisms and predators.
76
In addition to aggregations of C. oculatus, I found that several different species of
organisms existed on A. fimbriata: for example, colonies ofbryozoans, various species of
amphipods and gastropods, and small fish. In studies conducted on A. fimbriatum from
the west coast ofVancouver Island, British Columbia, Canada, some pieces of the kelp
blade were completely covered (at least on one side) by bryozoans (e.g., Lichenopora
novaezelandiae orMembranipora membranáceo) (Hurd et al. 1994). Previous studies
suggest that bryozoans influence the nitrogen (N) and photosynthetic abilities of the kelp
(Durante and Chia 1991, Hurd et al. 1994, but see Hurd et al. 2000). In some cases the
epiphytic stenolaemate bryozoans limit the ability of the kelp to remove nitrates and
ammonium from seawater but provide a source of ammonium to the kelp via excretion
(Hurd et al. 1994). According to field surveys by Durante and Chia (1991), bryozoan age
distributions are spatially correlated with algal age distributions; younger bryozoans were
concentrated on the youngest parts of the kelp (typically near the base), and vice versa.
Despite bryozoans limiting chemical uptake for their algal host, adult L. novaezelandie
has been found to deter gastropod feeding (e.g., herbivorous snail Tegula pulliog) on
older growth of kelp, revealing an indirect benefit of epiphytic bryozoans (Durante and
Chia 1991).
Younger bryozoan settlement on early growth ofA. fimbriata (Durante and Chia
1991) may provide insight for reasons that C. oculatus might be found primarily on new
growth (typically the base of the blade). I found no evidence that suggests younger
bryozoans limit feeding of gastropods and therefore, may be an indication that younger
bryozoans do not limit feeding for C. oculatus either. It is possible the older bryozoans
77
possess deterrents for C. oculatus and as a result, large numbers of individuals do not
aggregate at the tip of the algal blade.
Laboratory Experiments
Spatial distributions and patterns of aggregates. H2: There is no difference in
mean nearest neighbor distance for densities ranging from 753 individuals-m' (500
individuals-0.664m'^) to 7,530 individuals-m'^ (5000 individuals-0.664m'^). I rejected
this null hypothesis. The trend in nearest neighbor distances in this laboratory
experiment was not fully consistent with results observed the field. I expected that as
abundances increased in the testing pools, the distribution of amphipods would be forced
into tighter spacing minimums, creating a uniform pattern of 1 cm. Most densities
followed my hypothesis, with the notable exception of the highest density, 5000
individuals-0.664 m'^. Results indicated that as densities increased from 500, 1000, and
2000 individuals-0.664 m' , nearest neighbor distances declined 38% and approached 1
cm. However, from 2,000 to 5,000 individuals-0.664 m' nearest neighbor distances
increased 27% to 1.83 cm. Testing pools containing 5000 individuals were not different
than testing pools containing 1000 individuals. The pattern of the aggregation at 5000
individuals was clumped (Holgate 1965). Further review showed that the increased
change in the trend of decreasing MNND in the testing pool containing 5,000 individuals
was not dependent on the time at which the image was taken in the testing pool (2 hr
versus 24 hr). However, it should be noted that the largest number of nearest neighbor
distances < 0.5 pixels was eliminated at this density (23.8% or 1,190 NND) compared to
78
other densities, and it is possible the increased MNND was an outcome of extensive
mating and cannibalism efforts for Chrompleustes oculatus (Table 1). The clumped
pattern of distribution suggests that mating and cannibalistic C oculatus clumped
together and thus created additional space in various areas for other opportunistic
individuals to disperse. The pattern of dispersion was clumped for testing pools
containing 500 individuals and 1000 individuals, which coincides with previous 0.1-m^
visual quadrats taken of various habitats. The pattern of dispersion in the testing pool
2
containing 2000 individuals was random, which contradicts previous 0.0065-m
photoquadrats taken on C. oculatus at high densities that suggests C. oculatus form a
uniform pattern.
The time at which images were taken of testing pools (2 and 24 hours) had an
effect on mean nearest neighbor distances at various densities. Smaller densities of C.
oculatus decreased MNND when images were taken at 24 hours. It is known that C.
oculatus were mating in pools and it is possible the highly variable and large decrease in
MNND is the result of individuals trying to locate one another in testing pools containing
500 and 1,000 individuals-0.664 m'^. Testing pools with higher densities showed an
increased trend in MNND for images taken at 2 and 24 hours. It is documented that
amphipods can recolonize within large populations in a relatively short period of time;
e.g., a period of hours or days (Pavia et al. 1999). It is suggested the increase in MNND
(from 2 hours to 24 hours) in testing pools containing 2,000 and 5,000 individuals-0.664
m‘^ is the result of rapid colonization of the larger densities of C. oculatus and most likely
79
occurred to form tighter spacing minimums to maximize space for all individuals in the
community.
Laboratory observations suggested that density manipulation experiments were
conducted during the mating season for Chromopleustes oculatus. Various cohorts could
be seen throughout the sampling season from May to July, with July being the warmest
period for the Pacific Northwest. Amphipods were observed to initiate pairing with the
dominant individual, presumably the male, grasping its partner. After contact, the male
would beat its pleopods against the female. This pleopod beating and many other
behaviors noted in testing pools were characteristic of those described by Borowsky
(1991) on the reproductive biology of gammarid amphipods.
Most gammaridean amphipods are dioecious. Since females do not store sperm, a
male must accompany a female each time copulation and fertilization occurs, and females
can reproduce many times in succession. Females have four sequential stages; (1) mate
location (probably via waterborne pheromones), (2) initiation of pairing (stimulated by
contact), (3) pairing (continues until the female molts), and (4) copulation (which occurs
shortly after molt). Reproduction is seasonal in most natural populations, but in the
laboratory under constant conditions females have been observed to continuouslymolt
and reproduce while males continue to inseminate females (Borosky 1991, Kaestner
1970). Females ovulate minutes after molt, and the eggs pass into a ventral brood pouch,
or marsupium.
Results of observations concerning the controversy of the existence of waterborne
pheromones in reproductive behaviors have been contradictory. In a study using
80
Gammarus pulex, Ducruet (1973) concluded waterborne pheromones were present and
males were attracted to inter- and intraspecific females, particularly conspecifics at all
molt stages. It was not possible to replicate Dueruet’s findings in a later study by
Hartnoll and Smith (1980).
Temperature and day lengths are two environmental factors that have been shown
to affeet reproduetion in specific amphipods. The time it takes for the eggs to pass into
the marsupium (oviposit) varies with latitude and species. Some speeies oviposit in the
fall and release their young in the spring; others oviposit in the spring and release their
broods within a few weeks (Steele and Steele 1975).
Results ofmy study also suggested that C. oculatus eould be cannibalistic.
Approximately 10-15 amphipods in each testing pool would die before the 24 hours of
testing ended. In testing pools only, I noted several cases of cannibalism where
amphipods would attack a single dead or dying individual. At later observation times
only the exoskeleton would remain of the injured individual. In addition to attacks on
injured individuals, larger amphipods would attack smaller amphipods if left in holding
pools together.
Cannibalism in several freshwater gammarid amphipod speeies is not uncommon
(Embody 1911, Sexton 1924, Sexton 1928, Clemens 1950, Jones 1951, Schmitz 1967,
Kostalos and Seymour 1976, Jenio 1980, Dick 1995, MacNeil et al. 1997). Cannibalism
has been reported in many laboratory studies 1) as a dietary staple during depleted
resources, 2) as predation on injured, diseased, and dead individuals, 3) when adults feed
on juveniles, and 4) as a result ofmating competition (larger males feed on smaller
81
males; MacNeil et al. 1999). I personally observed larger, dominant amphipods attacking
smaller, injured, or dying amphipods in testing pools, in addition to noting several
amphipod remains in pools at testing times (molts not included). To my knowledge, this
is the first observation of cannibalism in a marine gammarid amphipod.
Predators and spatial distributions. H3: There is no difference in the mean
nearest neighbor distance in the presence ofpredators in the laboratory. This null
hypothesis was rejected. I proposed that Chrompleustes oculatus would decrease MNND
as defense in the presence of a predator, forming tighter spacing minimums. The
opposite held true for this series of experiments. The results indicated amphipods created
a significant increase in overall MNND in the presence ofpredators at a constant density.
Palatable amphipods are typically an important prey item (Kaiser et al. 1992, Pavia, Carr,
and Aberg 1998, Thiel and Reise 1993), but the chemically defended amphipod, C.
oculatus, is not a major source of diet for small fish and other predators (Hutchinson
2002, Zalewski 2001). Observations from Bell Island showed several species of small
predatory fish (i.e., sculpins) coexisting with C. oculatus. Further observations in the
laboratory indicated C. oculatus did not appear to be threatened by a naive predator
placed in testing pools.
It is possible the increased MNND was a result of eliminating nearest neighbor
distances < 0.5 pixels, leaving fortuitously higher mean distances. It cannot be ruled out
that the unique distributions seen in testing pools containing predators were a result of
mating or cannibalism, although the fact that the percentage of amphipods coming into
physical contact was reduced in the presence ofpredators (Table 1). The change in
82
distribution may have stemmed from the aggression that was seen from C. oculatus when
a single Oligocottus maculosus was placed in the tank. Individuals that attacked O.
maculosus increased space throughout the remaining testing pool, allowing amphipods to
disperse and increase MNND. One could argue that this latter scenario lends support to
the idea that amphipod warning coloration, which advertises their noxious chemicals, is
irrelevant in the presence ofpredators. It is also worth mentioning the p value for this set
of experiments {p = 0.05) was near the experiment-wise error rate and the number of
images used in calculations in the presence (n = 390) and absence (n = 50) of predators
may have biased the results in favor of a slight increase in MNND in the presence of
predators.
Habitat choice experiments. H5: The response of amphipods to habitat choices
will be the same for each of the four habitats: sea cucumbers, sand, rock, and sand-shell
debris. When presented with multiple habitats in the absence ofpredators,
Chromopleustes oculatus made distinct environmental choices. Bell Island was very
diverse in its habitats. Agarum algal blades were dominant at the northwest comer of
Bell Island, while a blanket of large starfish (Pycnopodia) up to 1 m in body diameter
covered the rocky terrain at the southeast portion of the island. Only 50 m west at
another rock outcropping known as Twin Peaks, a blanket of sea cucumbers (Cucumaria
miniata, Eupentacta quinquesemita, and Parastichopus californicus, to name a few)
existed on rock formations in addition to burrowing their trunks into the sand and shell
debris. Amphipods flourished in all these habitats, although they were not as dense with
the Pycnopodia along the southeast comer of Bell Island. When similar choices were
83
presented in the laboratory, C. oculatus actively inhabited C. miniata in all aquaria. It
was especially notable that C. miniata occupied the smallest two-dimensional area for
both aquarium A and B.
Evidence presented in Hutchinson (2002) suggests it is possible that selection for
C. miniata was directly related to foraging attempts. It is not known how C. oculatus
feeds, or what comprises its diet. Crustaceans in general have exploited many
combinations of feeding strategies associated with suspension feeding, filter feeding, or
direct feeding by manipulating mouthparts. Gammaridean amphipods are commonly
referred to as “sand grazers” and “sand lickers,” indicative of feeding on benthic diatoms,
detritus, or microorganisms located on surfaces of sediment particles (Brusca and Brusca
1990). In the study using diet analysis and laboratory behavior to determine if C.
oculatus sequesters its defenses through endogenous or exogenous mechanisms,
Hutchinson (2002) found the amphipod’s diet comprised largely of sea cucumber
ossicles, most likely belonging to Cucumaria miniata or Cucumaria piperata. In
addition to cucumber ossicles, echinoderm pedicellaria, centric diatoms, pinnate diatoms,
and other crustaceans (although not a major portion of identifiable prey) were found in
fecal matter (Hutchinson 2002).
It cannot be ruled out that amphipods may have selected sea cucumbers for other
reasons not associated with cucumbers providing food (either directly or indirectly as a
result of opportunistic amphipods intercepting the food during filter-feeding by the
cucumber). It is possible that sea cucumbers, when elongated, may have presented the
largest three-dimensional area compared with other habitats that presented no relief
84
structures. Aerial views of the aquaria limited analysis to two dimensions. Additionally,
three-dimensional views were not practical for this study due to the live nature of the sea
cucumber: the sea cucumber would shift positions throughout the 24 hour observations,
sometimes feeding with its branched tentacles or other times it would be found in a
resting state with its tentacles retracted.
C. oculatus possess multiple deterrents against predation including, but not
limited to, chemical defenses, aposematic coloration, gregariousness, and aggressive
behaviors. It has been shown that predatory fish in the lab will reject Chromopleustes
oculatus after previous encounters, suggesting a learning capability possibly in
conjunction with the brightly-colored dispositions and chemical defenses (Hutchinson
2002, Zalewski 2001). The hooded nudibranch, Melibe leonina, also does not appear to
be a potential predator. In a laboratory environment, I personally observed Melibe to
“regurgitate” individuals of C. oculatus when swallowed and to fully consume palatable
amphipods {Allorchestes spp.). For the invertebrate, there were no avoidance lessons
learned from bright coloration and distastefulness since Melibe are absent of sight.
The source of chemical defense in C. oculatus has recently been studied by
Hutchinson (2002). Hutchinson (2002) suggests that as adults, defenses are synthesized
de novo instead of sequestered from prey, regardless of diet. Starved adults that had their
defenses removed were just as defended as those provided with food after a period of six
days. Hutchinson’s (2002) study also showed that eggs taken from the marsupium were
palatable and eaten by all predators in the lab. Newly released juveniles did not show a
statistically different consumption rate in comparisons of palatable prey, but they were
85
rejected in 20% of trials with predators. Despite the lack of significance in Hutchinson’s
work (2002), this is other evidence that some juveniles are indeed defended inside the
marsupium. Further research on whether juveniles obtain defenses from the adult or
create them endogenously is warranted.
One could argue that the field observations from this study supported the concept
that aggregation in defended species reinforces chemical defenses (Hamilton 1971,
Mathews 1977). Amphipods created dense populations within a small area in the field as
a result of decreased spacing. Previous research has not indicated any fish predators for
this species (Zalewski 2001, Hutchinson 2002) and it is possible that a swarm of
individuals creates a toxic aversion.
Chemical defenses did not play a major role in spacing among individuals.
Densities in the laboratory followed similar spacing distributions. However, these density
changes were solely among conspecifics and not in the presence ofpredators. In
addition, further investigations in the laboratory in the presence of a potential predator
showed that chemical defenses were not reinforced by aggregations, suggesting other
biological factors were at work. Recall C. oculatus increased MNND and moved away
from conspecifics in the presence of predators.
Hutchinson (2002) states that C. oculatus does not sequester its defensive
compound but produces it de novo and therefore aggregations of individuals are not the
result of a need to forage on a defended host (e.g., A. fimbriatum or C. miniata). Various
habitats presented in the laboratory in the absence ofpredators further suggest that
chemical defense is not a driving force in habitat choices for C. oculatus.
86
Further studies on various vertebrates (fish) and invertebrates (e.g., predatory
nemerteans) would be beneficial in determining if these amphipods are consumed by any
predators. Perhaps C oculatus is influenced by other predators that have not been
determined. MacNeil et al.’s review (1999) highlights the dangers of focusing only on
conventional predators of Gammarus spp. such as fish used in this study and other works
cited in this paper. It is equally dangerous to assume Gammarus spp. have ecologically
equivalent behavior patterns, physiochemical tolerances, invasive potentials, and most
importantly for this study, resistance to invasion and predatory tendencies (MacNeil et al.
1997).
Subsequent work based on the observations that amphipod numbers are highly
variable on algal blades at slack tide would be useful in determining factors for mobility.
What water velocities can amphipods withstand before they are swept off an Agarum
blade? This is a key question in understanding amphipod activity in the field and must be
tested in a laboratory environment. With high variability in abundance for each algal
blade at different tidal states, in addition to strong currents at Bell Island, it would be
surprising if C. oculatus were able to create specific domiciles on individual blades
without being detached by the flowing waters of the exchange. Additional density
studies should focus on research created during the non-mating season for
Chromopleustes oculatus. This would address whether or not the inconsistencies in
experiments manipulating densities and the results from the presence ofpredators were
errors that stem from the distances removed from calculations due to mating and
cannibalism.
CONCLUSIONS
A variety of factors influenced the spatial distribution of the chemically defended,
aposematic, gammarid amphipod, Chromopleustes oculatus in the field and laboratory.
Amphipod location on Agarum blades affected the distribution ofnatural populations.
Large numbers of individuals could be found at the base of the blade but numbers were
reduced at the center and tip. As abundances increased on Agarum fimbriatum in the
field the mean nearest neighbor distances decreased, reflecting avoidance minimums
established by the community. Density, predators, and available habitats affected the
spatial distribution of C. oculatus in the laboratory. The trend observed forMNND in the
field (as density increases MNND decreases) was not seen for all densities in experiments
conducted in the laboratory. As density increased in testing pools containing 500 to
2,000 individuals-0.664 m' MNND decreased to establish minimum spacing among
individuals. Further review of the testing pools containing 5000 individuals indicated
this density had the highest percentage of nearest neighbor distances eliminated from
calculations than any other density as a result ofmating. The liberal numbers of
amphipods mating in pools created additional space for other individuals and are most
likely the result for the increase in MNND. Contrary to expectations, the MNND in the
presence of a naive fish predator was significantly higher than the MNND in the fish’s
absence. It possible this increase is the result ofpredator avoidance strategy by the
amphipods. Habitat significantly affected spatial distributions of amphipods. A higher
percentage of amphipods per area aggregated on the sea cucumber, Cucumaria miniata
despite various other substrate choices. Mean nearest neighbor distances that did not
88
coincide with predicted trends were presumably the result ofmating, cannibalism, or
predator avoidance strategies within amphipod communities.
A variety of abiotic and biotic factors influenced the spatial distribution of C.
oculatus such as density, proximity of amphipods to the base of algal blades, the presence
ofpredators, and habitat availability. However, it is my beliefs that this study suggests
chemical defenses only play a small role for Chromopleustes oculatus. Chemical defense
did not appear to regulate spacing in the presence ofpredators in the laboratory; tighter
spacing to reinforce toxins was not seen. Other laboratory experiments that evaluated
aggregation as a function of density changes and habitat choices were conducted without
predators and further suggests that aggregations to reinforce chemical defenses are not an
elementary driving force for C. oculatus.
LITERATURE CITED
Aikins, S. and E. Kikuchi. 2001. Studies on habitat selection by amphipods using
artificial substrates within an estuarine environment. Hydrobiologia 457: 77-86.
Aguilar-Perera, A. and W. Aguilar-Davila. 1996. A spawning aggregation ofNassau
grouper Epinephalus striatus (Pisces: Serranidae) in the Mexican Caribbean.
Environmental Biology of Fishes 45:351-361.
Alatalo, R.V. and J. Mappes. 1996. Tracking the evolution ofwarning signals. Nature
382:708-709.
Anderson, P.K. 1997. Shark Bay dugongs in summer. l:Lek mating. Behavior
134:433-462.
Anderson, R.S. 1970. Predator-prey relationships and predation rates for crustacean
zooplankters from some lakes in western Canada. Canadian Journal Of Zoology
48: 1229-40.
Borowsky, B. 1991. Patterns ofReproduction of Some Amphipod Crustaceans and
Insights into the Nature ofTheir Stimuli. Pages 33-49 in R. Bauer and J. Martin,
(eds.). Crustacean Sexual Biology. Columbia University Press, New York.
Bowers, M.D. 1993. Aposematic caterpillars: Life-Styles of the Wamingly Colored and
Unpalatable. Pages 331-371 in N.E. Stamp and T.M. Casey (eds.). Caterpillars.
Ecological and Evolutionary Constraints on ForaRing. Chapman and Hall, New
York.
Brower, J.E., Zar, J.H., and C.N. von Ende. 1998. Population dispersion. Pages 158-159
in Field and Laboratory Methods for General Ecology, 4^^ edition.
WCB/McGraw-Hill Publishers, New York.
Brusca, R.C., and G.J. Brusca. 1990. Invertebrates. Sinauer Associates, Inc. Publishers,
Massachusetts.
Carter, J., and D. Behrens. 1980. Gastropod Mimicry by another Pleustid Amphipod in
Central California. The Veliger 22(4): 376-377.
Clemens, H.P. 1950. Iáíq cycXt dtná ecology oí Gammarus fasiatus So-y. Contributions,
Stone Laboratory Ohio University. 12: 1-63.
Colinvaux, P.A. 1973. Introduction to Ecology. Wiley Publishing, New York.
90
Coma, R. and H.K. Lasker. 1997. Effects of spatial distribution and reproductive
biology on in situ fertilization rates of a broadcast-spawning invertebrate.
Biological Bulletin 193:20-29.
Crane, J. 1969. Mimicry of the gastropod Mitrella carinata by the amphipod Pleustes
platypa. Veliger 12:200.
David K., Barnes, A., and P. Rotherya. 1996. Competition in encrusting Antarctic
bryozoan assemblages: outcomes, influences and implications. Journal of
Experimental Marine Biology and Ecology 196(1-2): 267-284.
Davis, C.S., Flierl, G.R., Wiebe, P.H., and P.J.S. Franks. 1991. Micropatchiness,
turbulence, and recruitment plankton. Journal ofMarine Research 49: 109-151.
Dick, J.T.A. 1995. The cannibalistic behaviour of two Gammarus species (Crustacea:
Amphipoda). Journal of Zoology, London. 236: 697-706.
Domier, M.L. and P.L. Colin. 1997. Tropical reef fish spawning aggregations: Defined
and reviewed. Bulletin ofMarine Science 60:698-726.
Domenici, P. and R.S. Batty. 1994. Escape maneuvers of schooling Clupea harengus.
Journal of Fish Biology 45:97-110.
Donnelly, K. 1978. Simulations to determine the variance and edge-effect of total
nearest neighbour distance. Pages in 91-95 in I.Hodder, editor. Simulation
methods in archaeology. Cambridge University Press, London, England.
Ducruet, J. 1973. Interattraction chez les gammares du groupe pulex (Crustacés
Amphipodes). Rev. Comp. Animal 7:313-322.
Duffy, J.E. and M.E. Hay. 1991. Food and shelter as determinants of food choice by an
herbivorous marine amphipod. Ecology 72(4): 1286-1298.
Duffy, J.E. and M.E. Hay. 1994. Herbivore resistance to seaweed chemical defense: the
roles ofmobility and predation risk. Ecology 75(5): 1304-1319.
Durante, K.M. and F.S. Chia. 1991. Epiphytism on Agarum fimbriatum: Can herbivore
preferences explain distributions of epiphytic bryozoans? Marine Ecology
Progree Series 77(2-3): 279-287.
Edmunds, M. 1990. The evolution of cryptic coloration. Pages in D.L. Evans and J.O.
Schmidt (eds.). Insect Defenses. SUNY Press, Albany, NY.
91
Elliot, J.M. 1971. Upstream movements ofbenthic invertebrates in a lake district
stream. Journal ofAnimal Ecology 40(1): 235-252.
Embody, G.C. 1911. A preliminary study of the distribution, food and reproductive
capacity of some freshwater Amphipods. International Revue der Gesamten
Hydrobiologie und Hydrographie Supplement 3:1-35.
Facey, D.E. and M.J. Van-Den-Avyle. 1987. Species profiles: Life histories and
environmental requirements of coastal fishes and invertebrates. North Atlantic
American eel. Biological Rep. of the U.S. Fisheries and Wildlife Service pp. 35.
Field, L.H. 1974. A description and experimental analysis of Batesian mimicry between
a marine gastropod and an amphipod. Pacific Science 28:439-447.
Fisher, R.A. 1958. The genetic theory ofnatural selection, 2"*^ ed. Dover Publishing,
New York.
Fox, L.R. 1975. Cannibalism in natural populations. Annual review of Ecology and
Systematics 6:87-106.
Goddard, J. 1984. Presumptive Batesian mimicry of an aeolid nudibranch by an
amphipod crustacean. Shells and Sea Life 16:220-222.
Graham, W.M., Pages, F., and W.M. Hamner. 2001. A physical context for gelatinous
zooplankton aggegations: a review. Hydrobiologia 451(1-3): 199-212.
Grudemo, J. and C. Andre. 2001. Salinity dependence in the marine mud snails
Hydrobia ulvae and Hydrobia ventrosa. Journal ofMarine Biological
Association UK. 81(4): 651-654.
Guilford, T. 1985. Is kin selection involved in the evolution ofwarning coloration?
Oikos45: 31-36.
Hamilton, W.D. 1971. Geometry for the shellfish herd. Journal of Theoretical Biology
31:295-311.
Harder, W. 1968. Reactions ofplankton organisms to water stratification. Limnology
and Oceanography 13(1): 156-168.
Hartnoll, R.G. and S.M. Smith. 1980. An experimental study of sex discrimination and
pair formation in Gammarus duebeni (Amphipoda). Crustaceana (Leiden) 38:
253-264.
92
Hopper, A.F. 1960. The resistance ofmarine zooplankton of the Caribbean and South
Atlantic to changes in salinity. Limnology and Oceanography 5(1): 43-47.
Holgate, P. 1965. Some new tests of randonmess. Journal ofEcology 53:261-266.
Hurd, C.L., Durante, K.M., Chia, F.S. and P.J. Harrison. 1994. Effect of bryozoan
colonization on inorganic nitrogen acquisition by the kelps Agarum fimbriatum
and Macrocystis integrifolia. Marine Biology 121: 167-173.Hurd, C.L. and C.L.
Stevens. 1997. Flow visualization around single- and multiple-bladed seaweeds
with various morphologies. Journal of Phycology 33(3): 360-367.
Hurd C.L., Durante, K.M., and P.J. Harrison. 2000. Influence of bryozoan colonisation
on the physiology of the kelp Macrocystis integrifolia (Laminariales, Phaeophyta)
from nitrogen-rich and -poor sites in Barkley Sound, British Columbia, Canada.
Phycologia 39 (5): 345-440.
Hutchinson, M.E. 2002. Chemical Defense in the Gammarid Amphipods
Chromopleustes oculatus, Chromopleustes lineatus and Cryptodius kelleri.
Master of Science thesis. Department ofBiology, East Carolina University.
Hynes, H.B.N. 1955. The reproductive cycle of some British freshwater Gammaridae.
Journal of Animal Ecology 24: 352-387.
Jenio, F. 1980. The life-cycle and ecology of Gammarus troglophilus. Hubricht and
Mackin. Crustaceana (Suppl. 6), 204-215.
Jones, J.R.E. 1951. An ecological study of the River Towy. Journal of Animal Ecology
18: 142-159.
Kaestner, A. 1970. Invertebrate Zoology, Crustacea. Wiley Publishing, New York.
Kaiser, M.J., Gibson, R.N., and R.N. Hughes. 1992. The effect ofprey type of the
predatory behaviour of the fifteen-spined stickle-back, Spinachia spinachia (L.).
Animal Behaviour 43: 147-156.
Kaartvedt, S., Van Dover, C.L., Mullineaux, L.S., Wiebe, P.H., Bollens, S.M. 1994.
Amphipods on a deep-sea hydrothermal treadmill. Deep Sea Research A
Occeanogr. Res. Pap. 41(1): 179-195.
Kingsford, M.J., Leis, J.M., Shanks, A., Lindeman, K.C., Morgan, S.G., and J. Pineda.
2002. Sensory environments, larval abilities and local self-recruitment. Bulletin
ofMarine Science 70(l):309-340.
93
Kinne, O. 1960. Gammarus salinus-emxgQ. Daten under den Umwelteinfluss auf
Wachstum, Hautungsfolge, Herzfrequenz und Eientwicklungsdauer. Crustaceana
1: 208-217.
Kostalos, M. and R.L. Seymour. 1976. Role ofmicrobial enriched detritus in the
nutrition of Gammarus minus (Amphipoda). Oikos 27: 512-516.
Krapp, S.G. 1993. Do algal-dwelling amphipods react to the "critical zones" of a coastal
slope? Journal ofNatural History 27(4): 883-900.
Krause, J. 1993. The effect of “schreckstoff’ on the shoaling behavior of the minnow: A
test ofHamilton’s selfish herd theory. Animal Behaviour 45:1019-1024.
Krebs, C.J. 1989. Ecological methodology. Harper & Row, New York.
Kureck, A. 1967. Uher die tagesperiodische Ausdrift von Mjo/iargMá'
schellenbergl Karaman aus Quellen. Z. Morph. Okol. Tiere 58: 247- 262.
Larkin, P.A. 1956. Interspecific competition and population control in freshwater fish.
Journal ofFisheries Research Board of Canada 13: 327-42.
Lehman, U. 1967. Drift und Populationsdynamik von Gammarus pulexfossarum Kock.
Z. Morph. Okol Tiere 60:121-21A.
Leising, A.W. and J. Yen. 1997. Spacing mechanisms within light-induced copepod
swarms. Marine Ecology Progress Series 155: 127-135.
Levitan, D. 1998. Does Bateman’s principle apply to broadcast-spawning organisms?
Egg traits influence in situ fertilization rates among congeneric sea urchins.
Evolution 52:1043-1056.
Levitan, D.R. and C. Petersen. 1995. Sperm limitation in the sea. Trends. Ecological
Evolution 10:228-231.
Ludwig, J.A, and J.F. Reynolds. 1988. Statistical ecology. A primer on methods and
computing. John Wiley and Sons, New York.
Macan, T.T. 1965. Predation as a factor in the ecology ofwater bugs. Journal of
Animal Ecology 34: 691-98.
Macan, T.T. and J.C. Mackereth. 1957. Notes of Gammarus pulex in the English Lake
District. Hydrobiologia 9:-l-12.
94
MacNeil, C., Dick, J.T.A., and R.W. Elwood. 1997. The trophic ecology of freshwater
Gammarus (Crustacea: Amphipoda); problems and perspectives concerning the
functional feeding group concept. Biological Reviews of the Cambridge
Philosophical Society 72: 349-364.
MacNeil, C., Dick, J.T.A., and R.W. Elwood. 1999. The dynamics of predation on
Gammarus spp. (Crustacea: Amphipoda). Biological Review UK. 74; 375-395.
Manzi, J.J. 1970. Combined effects of salinity and temperature on the feeding,
reproductive, and survival rates ofEupleura caudata (Say) and Urosalpinx
cinerea (Say) (Prosobranchia : Muricidae). Biological Bulletin 138: 35-46.
Maravelias, C.D. 1997. Trends in abundance and geographic distribution ofNorth Sea
herring in relation to environmental factors. Marine Ecology Progress Series
159:151-164.
Mathews, E.G. 1977. Signal-based frequency-dependent defense strategies and the
evolution ofmimicry. American Naturalist 111 :213-222.
McCartt, A., Lynch, W., and D. Johnson. 1997. How light, a predator, ad experience
influence bluegill use of shade and schooling. Environmental Biology of Fishes
49: 79-87.
Meade, M.E., Doeller, J.E., Kraus, D.W., and S.A. Watts. 2002. Effects of temperature
and salinity on weight gain, oxygen consumption rate, and growth efficiency in
juvenile red-claw crayfish Cherax quadricarinatus. Journal of the World
Aquaculture Society 33(2): 188.
Minckley, W.L. 1964. Upstream movements ofGammarus (Amphipoda) in Doe Rin,
Meade County, Kentucky. Ecology 45(1): 195-197.
Mottram, J.C. 1933. Some observations on the life history of the freshwater shrimp.
Transactions of the Newbury District Field, Club. 7:2-4.
Muller, K. 1954. Investifations on the organic drift in north Swedish streams. Rep. Inst.
Freshwat. Res. Drottningholm 35: 133-148.
Mueller, K.W. 1994. Gregarious behavior in mutton snapper in the Exuma Cays.
Bahamas Journal of Science 1:17-22.
Muspratt, J. 1951. The bionomics of an African Megarhinus (Dipt., Culicidae) and its
possible use in biological control. Bulletin of Entomology Research 42: 355-70.
95
Nikolskii, G.V. 1969. Theory of Fish Population Dynamics. Oliyer and Boyd,
Edinburgh.
Norton, S.F. and C.D. Stallings. 1999. The distribution and abundance of the aposematic,
chemically defended gammarid amphipods off Bell Island, WA. J.N. Heine, Don
Canestro, G. Wuttken (eds.), Diying for Science 1999. Proceedings of the
American Academy ofUnderwater Sciences.
Paine, R.T. 1965. Natural history, limiting factors and energetics of the opisthobranch
Navanax inermis. Ecology 46: 603-19.
Parker, S.J. 1995. Homing ability and home range of yellow-phase American eels in a
tidally dominated estuary. Journal of the Marine Biological Association U.K.
75(1): 127-140.
Parish, J.K. 1989. Re-examining the selfish herd: Are central fish safer? Animal
Behaviour 38:1048-1053.
Pavia, H., Carr, H. and P. Aberg. 1998. Habitat and feeding preferences of crustacean
mesoherbivores inhabiting the brown seaweed Ascophyllum nodosum (L.) Le Jol.
and its epiphytic macroalgae. Journal of Experimental Marine Biology and
Ecology 236: 15-32.
Poore, A.G.B., Watson, M.J., de Nys, R., Lowry, J.K., and P.D., Steinberg. 2000.
Patterns of host use among alga- and sponge- associated amphipods. Marine
Ecology Progress Series 208: 183-196.
Prince, J.D. and D.A. Griffin. 2001. Spawning dynamins of the eastern gemfish {Rexea
solandri) in relation to regional oceanography in south-eastern Australia. Marine
Freshwater Research. 52(4): 611-622.
Rosenberg, G. 1989. Aposematism evolves by individual selection: Evidence from
marine gastropods with pelagic larvae. Evolution 43:1811-1813.
Sargent, T.D. 1990. Startle as an anti-predatormechanism, with special reference to the
underwing moths, (Catocala). Pages 227-249 in D.L. Evans and J.O. Schmidt
(eds.). Insect Defenses. SUNY Press, Albany, NY.
Sauer, W.H.H., Roberts, M.J., Lipinski, M.R., Smale, M.J., Hanlon, R.T., Webber, D.M.,
and R.K. O’Dor. 1997. Choreography of the squid’s “nuptial dance”. Biology.
Bulletin. 192:203-207.
Schmitz, E.H. 1967. Visceral anatomy of Gammarus lacustris lacustris Sars (Crustacea:
Amphipoda). American Midland Naturalist 78: 1-54.
96
Sexton, E.W. 1924. The moulting and growth-states of Gammarus, with descriptions of
the normal and intersexes of G. chevreuxi. Journal of the Marine Biological
Association UK. 13: 340-341.
Sexton, E.W. 1928. On the rearing and breeding of Gammarus in laboratory conditions.
Journal of the Marine Biological Association UK 15: 33-35.
Sillen-Tullberg, B., and O. Leimar. 1988. The evolution of gregariousness in distasteful
insects as a defense against predators. American Naturalist 132:723-734.
Sotka, E.E., Hay, M.E., and J.D. Thomas. 1999. Host-plant specialization by a non-
herbivorous amphipod: advantages for the amphipod and costs for the seaweed.
Oecologia 118: 471-482.
Stamp, N.E. and R.T. Wilkens. 1993. On the cryptic side of life: being unapparent to
enemies and the consequences for foraging and growth in caterpillars. Pages 283-
330 in N.E. Stamp and T.M. Casey (eds.). Caterpillars. Ecological and
Evolutionary Constraints on Foraging. Chapman and Hall, New York.
Steele, V.J. and D.H. Steele. 1973. The biology ofGammarus (Crustacea; Amphipoda)
in the nothwestem Atlantic, VII. The duration of embryonic development in five
species at various temperatures. Canadian Journal ofZoology 51: 995-999.
Steele, V.J. and D.H. Steele. 1975. The biology of Gammarus in the northwestern
Atlantic. XL Comparison and discussion Canadian Journal of Zoology 53: 1116-
1126.
Straty, R.R. 1972. Ecology and behavior a juvenile sockeye salmon (Oncorhynchus
nerka) in Bristol Bay and the eastern Bering Sea. Proceedings of the International
Symposium for Beri Hood, D.W. pp. 285-319.
Styan, C.A. 1998. Polyspermy, egg size, and the fertilization kinetics of tree-spawning
marine invertebrates. American Naturalist 152: 290-297.
Takeuchi, I. and K. Watanabe. 1998. Respiration rate and swimming speed of the
necrophagous amphipod Eurythenes gryllus from Antarctic deep waters. Marine
Ecology Progress Series 163: 285-288.
Thiel, M. and K. Reise. 1993. Interactions of nemertines and their prey on tidal flats.
Netherlands Journal of Sea Research 31, 163-172.
Treisman, M. 1975. Predation and the evolution of gregariousness. I. Models for
concealment and evasion. Animal Behaviour 23:779-800.
97
Tullrot, A. 1994. The evolution of unpalatability and warning coloration in soft-bodied
marine invertebrates. Evolution 48:925-928.
Turner, G.F., and T.J. Pitcher. 1986. Attack abatement: a model for group protection by
combined avoidance and dilution. American Naturalist 128: 228-240.
Uye, S. and A. Fleminger. 1977. Effects of various environmental factors on egg
development of several species of Acartia in Southern California. Marine
Biology 38(3): 253-262.
Van Dolah, R.E. 1978. Factors regulating the distribution and population dynamics of
the amphipod Gammarus palustris in an intertidal salt marsh community.
Ecological Monographs. 48(2): 191-217.
Vehrencamp, S.L. and J.W. Bradbury. 1984. Mating systems and ecology. Pages 251-
278 in Krebs and Davies (eds.), Behavioural Ecology: An evolutionary approach.
Vulinec, K. 1990. Collective security: aggregation by insects as a defense. Pages 251-
288 in D.L. Evans and J.O. Schmidt (eds.). Insect Defenses. SUNY Press,
Albany, NY.
Welton, J.S. and R.T. Clark. 1980. Laboratory studies on the reproduction and growth of
the amphipod, Gammarus pulex (L.). Journal ofAnimal Ecology 49(2): 581-592.
Wippelhauser, G.S. and J.D. McCleave. 1987. Precision of behavior ofmigrating
juvenile American eels {Anguilla rostrata) utilizing selective tidal stream
transport. J. Conser. Ciem. 44(1): 80-89.
Zalewski, J.A. 2001. Predatory-prey interactions between three chemically defended,
aposematic gammarid amphipods and cottid fishes. Master of Science thesis.
Department ofBiology, East Carolina University.