PRIOR EXPERIENCE AS A FACTOR IN
HABITAT SELECTION OF THE
CICHLID FISH TILAPIA MOSSAMBICA

Thesis for the Degree of Ph. D.
MICHIGAN STATE UNIVERSITY
CARL JOHN QUERTER'MUS, JR.

1972

   
     

  

LI B R“ " V
_ Michigan barre
University

w".

This is to certify that the

thesis entitled

Prior Experience as a Factor in Habitat Selection
of the Cichlid Fish Tilapia mossambica

presented by

Carl John Quartermus, Jr.

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Zoology

WGWWQ

quor professor

 

Date July 12, 1972

0-7639

; mamas! _
BOAT; & SUNS'
800K BINDERY INC. 5

LIBRARY BINDERS

Gill-nun" mun-nu

   
 

 

 

 

 

ABSTRACT

PRIOR EXPERIENCE AS A FACTOR IN HABITAT SELECTION
OF THE CICHLID FISH TILAPIA MOSSAMBICA

By

Carl John Quertermus, Jr.

An African Cichlid fish, Tilapia mossambica (Peters) was selected
to investigate the role that prior habitat experience may play in
determining the choice of a resting habitat in fish. Fish were
"conditioned" (reared) to different habitats and later tested for habitat
selection in an experimental tank offering a choice of the conditioned
and an alternative habitat. Three regimes of experience were used (early,
late, and continuous). Early-experience fish received experience with the
conditioned habitat the first 60 days of life, late-experience fish the
laSt 60 days before testing, and continuous-experience fish from the time
of free-swimming to testing. Within each regime, 20 fish conditioned to
a habitat with a sand substrate and 20 conditioned to a habitat with a
stone substrate were tested. Half of the fish were tested at four
months and half at eight months. Each individual was observed for one
hour during a two-day period in the experimental tank. This tank
offered a choice of two sand and two stone habitats.

Early and continuous-experience fish (both four and eight months)
significantly selected the conditioned habitat on the second day in the
testing tank in terms of total time and number of habitat visits. There
was no significant difference between fish given early and continuous
experience. There was an age difference for late-experience fish on day

two. Four-mmnth fish showed no significant effect of prior experience,

Carl John Quertermus, Jr.

but eight-month fish strongly avoided the conditioned habitat on day two.
Possible explanations for this avoidance are discussed. It is concluded
that early experience (before sexual maturity) is not only a sufficient,
but also a necessary, condition for habitat fixation in Tilapia
mossambica as judged by a short term selection of two days. These
results are discussed in relation to the natural history of g, mossambica
and the possible application of this study to fisheries management.

The relationship of stress and habitat selection is discussed in
terms of the findings that: (1) selection was made on day two but not
on day one; (2) fish that fed on day two had a relatively strong
selection for the conditioned habitat; and (3) color patterns indicating
fright were relatively'more common for fish visiting the unconditioned
habitat.

Several behavioral comparisons were made using swim time, number of
swim.bouts, number of pivots, and number of substrate manipulations.
There were no definite differences in activity scores for the
comparisons of (1) day one vs. day two and (2) fish in conditioned
habitat vs. fish in unconditioned habitat. Stone-conditioned fish did
more swimming (time and bouts) and.more substrate manipulating than sand-
conditioned fish. However, fish (both sand and stone-conditioned)
visiting the sand habitat did more swimming (time and bouts) and.more
substrate manipulating than fish in the stone habitat. Likewise, four-
month fish did more swimming (time and bouts) and more substrate
manipulating than eightqmonth fish. Pivoting scores were similar or
identical for all comparisons, which suggests that this behavior is
unaffected by experience, age, or habitat. Pivoting may be necessary
for a fish to maintain its "bearings" or awareness. The results of the

behavioral comparisons are discussed in relation to other studies.

PRIOR.EXPERIENCE AS A FACTOR IN HABITAT SELECTION
OF THE CICHLID FISH TILAPIA MOSSAMBICA

By

Carl John Quertermus, Jr.

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Zoology

1972

IQCOpyright by
Carl John Quarterms, Jr.
1972

To my daughters, Kristin and Lisa

ii

ACKNOWLEDGMENTS

I greatly appreciate the support and guidance of Dr. James C.
Braddock, who served as my major professor. I also wish to thank
Dr. Eugene Roelofs and Dr. Clarence McNabb from the Fisheries and
Wildlife Department and Dr. William Cooper from the Zoology Department
for serving as committee members. Dr. Cooper kindly helped with the
statistical analysis.

I am most thankful for the help, suggestions, and encouragement
from my good friend Chris Thoms. The kind consideration and help of
the Zoology Department secretaries and Mr. James Lucas are greatly
appreciated.

W graduate work was made possible by the hard work and
understanding of m wife, Barbara and the support of our parents,

Mr. and Mrs. R. W. Caauwe and Mr. and Mrs. C. J. Quertermus.

iii

TABLE OF CONTENTS

LIST OF TABLES ....................................................
ESTOFFKWME..u.n.n.n.u.n.n.uuu.u.u.u.u.u.n.n..
nflmmmflflm..u.n.u.n.u.u.n.u.n.n.u.n.n.u.u.u.n..
NATURAL HISTORY OF TILAPIA MOSSAMBICA ..
Distribution .................................................
Ecology and Behavior .........................................
Economic and Scientific Value ................................
MMWSMDMEMMS"nunnnuununuuunnuunnnu.
Species Selection ............................................
Selection of a Habitat Parameter for Manipulation ............
Habitat Conditioning .........................................
Experimental Design ..........................................
Experimental Testing Tank ....................................
Procedures for Testing .......................................
Statistical Analysis .........................................
Immmfiiu.n.u.n.u.n.u.n.u.u.u.u.u.n.u.u.u.u.u.u
Effect of Conditioning .......................................
Effect of Age at Experience ..................................
Behavior Analysis ............................................

COlor Pattern Analij-s 00......00....OOOOOOOOOOOOOOOOIOOCOO...

iv

Page

vii

\IO\O\l-'

Page
Importance of Early Experience ... 50
Comparison of Days One, Two, and Two F . 54
Activity and Color Pattern ................................... 55

l. mossambica as a Representative Fish to Illustrate
Habitat selection O.........0.0.0.0.........OOOOOOOOOOOOOOOOOO 58

Applied Aspects of This Research ..... 59
HindSight and the mtwe 00............OOOOO.......COOOOOOOOOO 61

SUMMARY ........................................................... 63

LIST OF REFERENCES 0.0.0.0...0.........OOOOOOOOOOOOOOOCO0.0.0.0.... 65

13.

11+.

LIST OF TABLES

Average percent time spent in the sand habitat on day one,
day two, and day two F (fish that fed on day two) ...........

Analysis of variance table for time spent in the sand
habitat on day two .............OOOOOIOOOOOOIO......OOOOOOOOO

Number of visits to sand and stone habitats on day one,
day two, and day two F (fish that fed on day two) ...........

Analysis of habitat visits by chi-square twoaway
independence.t65ts 000......0.00.0.0........OOOOOOOOOOO......

Average percent time spent in conditioned habitat on day
one, day two and day two F (fish that fed on day two) .......

Analysis of variance table for time spent in the
conditioned habitat on day tun ............OOOCOCOOOOOOOOOOO.

A priori individual degree of freedom comparisons for
experience treatment (factor A) for day two time in
conditioned habitat 0.00.....00......OOOOOOOOOOOOOOOOOOOOO...

Summary of time variable for all groups with sand or stone
experience during the first two months (groups 1-4 and

9-12)

Activity comparison of fish on day one and day two.
Numbers in parenthesis indicate number of fish more active
on the indicated day than on the alternate day ..............

Behavior comparison of fish in familiar and unfamiliar
habitats for groups 1-# and 9-12 (day one and day two) ......

Behavior frequency comparison for behaviors expressed in
the sand and Stone habitats (all grOuPS) oooooooooooooooooooo

Behavior comparison of sand and stone conditioned fish for
all groups on day one and day two combined ..................

Behavior frequency comparison of fouramonth and eight-
month fish for all experimental groups combined .............

Comparison of coloration of fish selecting conditioned and
unconditioned habitats for groups 1.u and 9-12 ..............

vi

Page

25

27

31

32

36

37

39

#0

#2

“3

#7

49

LIST OF FIGURES

Figure

marmantal deSj-gn .0000...00.000.000.0000000000000000000000
Bank or three experimental tanks ......OCOO...OOOOOOOOOOOOOOO

Top view of an experimental tank showing location of
doorway and different habitats ......OOOOOOOOO......OOOCOO...

Graphic presentation of second-order interaction for time
in the sand habitat 0..........OOOOOCOOOOO.....OOOOOOOOOOOOOO

vii

Page

16
18

18

28

INTRODUCTION

Naturalists have long recognized that each animal species occupies a
particular habitat or is associated with particular habitat parameters.
Correlation of various habitat features and the distribution of fish
species has received considerable attention (e.g. Forbes, 190?; Hallam,
1959; Norris, 1963; Munther, 1970; Everest, 1969; Braasch and Smith,
1965; Jenkins, 1969; and Tyler, 1971), but only'a few studies have been
thorough or have considered a whole community of species (e.g. Day and
Pearcy, 1968; Welcomme, 1964; and Russell, 1969). Day and Pearcy'(1968)
studied the benthic species on the continental shelf and slope off Oregon
and.made correlations of species complexes with water depth and sediment
composition. 'Welcomme (l96fi) correlated the distribution of the young
of various Lake Victoria Tilapia with water temperature, depth, dissolved
oxygen, and distance from shore. Russell (1969) did a correlation
analysis of various physical features of the habitat and the distribution
of 62 marine, rocky, shore species of New Zealand.

These correlation studies suggest that any one species utilizes
only certain habitats. Therefore, fish must discriminate between
habitats and select among the variety available to them. Correlation,
however, does not necessarily'imply selection (Klopfer, 1969). In order
to determine whether fish are actually selecting a habitat it is
necessary to test them.under controlled conditions where a variety of
habitats are available. If a species spends significantly'more time in
one of the available habitats, a selection is being made.

1

In recent years various species of fish have been tested to
determine their habitat preferences. Baker (1971) observed that 94% of
the fourspine sticklebacks in a freshawater stream were restricted to a
habitat comprising about 1% of the total stream study area. The
distinguishing characteristic of this habitat appeared to be the
presence of Elggga canadensis. In a choice situation in the laboratory
the sticklebacks selected an Elgggg.habitat over a Potamogeton habitat,
which was also available in the field. In a more intensive study, Sale
(1968 and 1969b) showed that juvenile manini, Acanthurus trioste 3,
select a habitat primarily on the presence of cover, substrate, algal
food, and suitable water depth. The presence of conspecifics and.water
movement are of minor importance, and the following are of slight or no
importance: light intensity, substrate color and texture, temperature,
salinity, oxygen content, and distance from shore. Several other studies
have dealt with various aspects of habitat selection in various species
of Lelia and Micropterus (Baldes and Vincent, 1969; McCrimmon and Kwain,
1966; Kwain and MbCrimmon, l967; Haines and Butler, 1969; Butler and
Hawthorne, 1968; and Hunsaker and Crawford, 196k).

As with other behavior patterns, considerable controversy'has been
raised regarding the basis for habitat selection. Is habitat selection
genetically determined or is it a learned response? Surprisingly few
studies on vertebrates have considered this question. Harris (1952)
utilizing the deermouse, Peromyscus maniculatus, was one of the first to
study this problem. He laboratoryaraised two subspecies of
2. maniculatus. One subspecies is found in a field habitat and the other
in a woodland habitat. As adults, both subspecies selected their

appropriate habitat without ever being exposed to it earlier in life.

These results suggest a genetic determination for habitat selection in
this species, but the possibility that learning might also be involved
was not ruled out since it was not tested. Klopfer (1963) also fbund
that habitat selection in Chipping Sparrows is at least partly determined
by'heredity. Hand-reared Chipping Sparrows selected normally preferred
pine foliage.

The possibility that learning (prior experience) may be of
importance in habitat selection of vertebrates has only recently'been
tested. 'Wecker (1963 and 196h)'worked with the prairie deermouse,
Permscus maniculatus M, and found that habitat selection in this
subspecies is determined primarily by heredity, although prior experience
with a field habitat can reinforce this "innate" preference. He also
found that early (prior) experience in other habitats cannot override
the normal affinity of the field stock for a field habitat. A
laboratory stock (12-20 generations), however, lost most of the
hereditary control and was much more influenced by prior experience with
a habitat.

Klopfer (1963, 1965, and 1967) and Klopfer and Hailman (1965) have
investigated the importance of prior habitat experience on subsequent
habitat selection in Chipping Sparrows and Blue Tanagers. Their results
indicate heredity is of’most importance, with learning having some
effect. For example, field-caught or hand-reared Chipping Sparrows
showed a preference for pine foliage over oak, but birds raised in the
presence of oak foliage showed a decreased preference for pine. However,
by four to six months the decreased preference disappeared. Sargent
(1965) found that previous nesting experience influenced subsequent

nestings in the Zebra Finch. Only'certain aspects of nesting seem to be

learned (nesting material and nest location) while others are not
apparently influenced by experience (substrate).

A recent investigation (Wiens, 1970) suggests prior experience is
of'most importance in habitat selection of Rana,aurg£a tadpoles.

Tadpoles raised in plain tanks (controls) and those raised in tanks with
the walls and bottom colored in a checkerboard pattern showed no
preference in a choice situation of a checkerboard or a striped
patterned habitat. But individuals raised in a tank with black and
white-striped walls and bottom had a strong preference for the striped
habitat.

‘With the possible exception of the numerous studies that have
verified the importance of olfactory imprinting of a spawning "habitat"
in Pacific salmonids (e.g. Hasler, 1966), there have been no studies
that have investigated the possible importance that prior experience may
play in fish habitat selection. However, there have been several studies
suggesting that various habitat features are learned (e.g. Aronson, 1951;
Breder, 1950 and 1954; Hasler, 1956 and 1966; Hasler andeisby, 1958;
and Sale, 1971). Aronson's (1951) study on the interridal gobiid fish,
Bathygobius sgporator, indicates that these fish must utilize learned
features of the habitat when jumping from.one tide pool to another during
low tide. Hasler's (1966) findings that white bass populations in Lake
Mendota return to the same spawning area year after year suggests
learning of one or more features of the habitat. Recently, Sale (1971)
observed that two populations of a coral-reef fish, Dasgyllus‘gggeggg,
were associated with two different habitats differing with regard to the
species of coral present. In a laboratory choice situation both

populations selected the coral type they were feund associated with in

the field. These field observations and the investigations of fish
learning ability by experimental psychologists of the 1960's

(e.g. Behrend and Bitterman, 1964; Otis and Cerf, 1963; and werboff and
Lloyd, 1963) have discredited the long established notion that fish
behavior is simply a product of their heredity.

With the preceding knowledge: (1) that at least some species of
fish do indeed select their habitats; (2) that prior experience is of
some importance in habitat selection of other vertebrates; and (3) that
fish are capable of learning, I designed an investigation to ascertain
the possible importance of prior habitat experience and subsequent
selection of a habitat. The experiment was designed not only to
question the importance of prior experience, but also to establish when
in the life of a fish such experience is necessary to establish a
preference for a habitat and at what age this experience is utilized in
selecting a habitat. I especially wanted to determine the role that early
experience might play. The studies of Wecker (1963) and Klopfer (1963)
indicate experience with a habitat early in the life of an animal may be
of‘importance in the later selection of a habitat. Also, investigations
with a wide variety of animals have shown that early'experience is of'
most importance in the establishment of their adult behavior patterns
and preferences (e.g. Beach and Jaynes, 1954; Newton and Levine, 1968).

In order to complete this investigation in a reasonable period of
time, a species of fish was selected that is predictably bred and easily
maintained in the laboratory and that matures at an early age. An
African cichlid, Tilapia mossambica (Peters), fulfills these requirements

and pilot work showed this was a favorable species to use for the study;

NATURAL HISTORY OF TILAPIA MDSSAMBICA

Distribution

Tilapia mossambica (family Cichlidae) is indigenous to the lower
reaches of river systems in East and South Africa from the lower Zambesi
River south to Algoa Bay (Fryer, pers. comm.; Jubb, 1961; Trewaves,
1966). COpley (1958), however, claims the distribution of the species
extends as far north as the Juba River, some 150 north of the Zambesi.
At the southern end of its distribution,'T.:mossambica is restricted to
estuaries (Allanson, Bok, and van Wyk, 1971).

.T.:mossambica has been introduced widely'in tropical and subtropical
waters. The first introduction outside Africa was to Java in 1939
(Chen, 1953). Since that time, introductions have been made to Malaya,
the Philippines, Thailand, West Indies, Taiwan, Japan, India, Formosa,
Ceylon, Pakistan, Trinidad, Borneo, Burma, Malacca, Sumatra, Puerto
Rico, many South Pacific islands including Hawaii, the United States,
and elsewhere (Chen, 1953; Chimitis, 1955; Swingle, 1960; Devambez,
1961»). In the United States the species is established in California,
Arizona, Florida, Texas, and a warm spring pond in Montana (St. Amant,
1966; Hoover and St. Amant, 1970; Lachner, Robins, and Courtenay, 1970).

'T.:mossambica has numerous common names (Crass, 1964; Piennar,
1968). In the United States it has been designated the mozambique
mouth-brooder (Bailey et al., 1970). Considerable confusion has reigned
over the classification of this species, anle.:mossambica is now
equivalent to‘T. (Chromis) yggglefeffer,IT. (Chromis) natalensis weber,
and T. arnoldi Gilchrist and Thompson (Trewaves, 1966). f_1‘_. mortimeri
from.the middle Zambesi is very closely'related to‘T, mossambica and the

two species have only recently been separated (Trewaves, 1966). A
6

taxonomical description of‘T, mossambica is found in Trewaves (1966),
Crass (1964), Piennar (1968), and Jubb (1961). A classification for the

entire Tilapia genus is found in Tdys van den Audenaerde (1968).

Ecolog 3951 Behavior

In its native African habitat.T.:mossambica primarily inhabits
rivers and estuaries but also occurs in lagoons and swampy lakes (Fryer,
pers. comm.; Jubb, pers. comm.; Piennar, 1968). In rivers the species
inhabits deep, sluggish pools with considerable marginal vegetation,
especially Phragmites and Typhg and other aquatic vegetation such as
Aponogeton and Potamogeton (Jubb, pers. comm.). In lakes it is sometimes
found in water more than 10 m deep (Boltt, pers. comm.).

The fish flourishes at water temperatures between 21 C and 26 C
(Jubb, pers. comm.). The lower lethal temperature is reported as 8 to
13 C depending on the authority (Jubb, pers. comm.; Crass, pers. comm.;
Beltt, pers. comm.; Piennar, 1968; Allanson et al., 1971). However, in
saline water this fish can tolerate temperatures to 5 C (Jubb, pers.
comm.). The lower tolerance in saline water apparently explains why'

.2. mossambica is restricted to estuaries at the southern part of its
distribution (Allanson et al., 1971). The upper lethal temperature is
38.2 C (Allanson and Noble, 1964).

Laboratory tests indicate a higher temperature preferendum.than
field studies. Badenhuizen (1967) found a mean temperature preference
of 28.5 C (27.0 to 33.5) in the lab as compared to 25 C reported by Jubb
(1961) in the field. Boltt (pers. comm.) mentioned that young fish have
temperature preferenda of a very high value which decreases as the fish
increase in size. He suggests that this enables the young to penetrate

into shallow pools with high temperatures (40 C) where they feed. His

data indicate that immature individuals inhabit shallower water than
sexually'mature individuals. This observation might be explained by'the
difference in temperature preference of immature and mature fish.

I. mossambica is euryhaline, tolerating salinities from 0 to 32%
(Jubb, pers. comm.). Crass (pers. comm.) mentioned that fish are found
in water with a salinity of 50%, but Jubb (pers. comm.) reported that
45% approaches the lethal limit. This remarkable fish is capable of
breeding in sea water (Brock, 1954), and there is a rumor that they are
established in the Pacific Ocean off Hawaii (Fryer, pers. comm)! Jubb
(pers. comm.) mentioned that I. mossambica can live in organically
polluted water with a dissolved oxygen content of‘3 ppm or less.

In native waters, breeding takes place during the spring (September
through December). Males move into shallower water - usually 3 to 6
feet (Jubb, 1967); less than 7 m in Lake Sibaya (Boltt, pers. comm.) -
where they construct a saucer-like nest a foot or more in diameter
(Crass, 1964). Jubb (1967 and pers. comm.) claims that a male digs and
defends several adjacent nests, but other authorities suggest only one
next is prepared. The nests are normally constructed in sandy
substrates, but muddy'substrates in both rivers and impoundments have
been utilized (Boltt, pers. comm.). Males prefer course sand, 2 to 3 mm
diameter (Jubb, pers. comm.), and hippopotamus tracks are often used as
nests (Piennar, 1968). The fish nest both in vegetation and below the
limits of vegetation (Boltt, pers. comm.).

The nest is made by mouth digging and is defended by the resident
male (Baerends and Baerends-Van Bean, 1950; Neil, 1964 and 1966). Males
are polygamous, and females move into the male territories and after

considerable courtship display, spawning begins (Jubb, pers. comm.).

3381

desc

the

Baerends and Baerends-Van Roon (1950) and Neil (1964 and 1966) have
described courtship, spawning, and parental behavior of this species in
the laboratory and Hawaiian ponds.

A female deposits a few eggs (about 10) in the pit of a male
(Baerends and Baerends-Van Roon, 1950) and then immediately picks them
up with her mouth. The male swims over the place where the eggs were
layed and ejects milt, which the female also takes into her mouth
(Baerends and Baerends-Van Room, 1950 and Neil, 1964 and 1966). Other
authorities indicate that the male emits his milt over the eggs before
the female picks them up (Chimits, 1955; Crass, 1964; Chen, 1953; Jubb,
pers. comm.). .After spawning the female moves off the breeding grounds
and usually selects a shallow backawater area (Jubb, pers. comm.). She
broods the eggs in her mouth. The eggs hatch in about 60 hours at 28 C
but remain in the female's mouth for another 5 to 8 days (Chen, 1953).
At the end of the larval period the young become free-swimming and
leave the mother's mouth for the first time. At this point they are
about 8 mm total length. The young remain with the mother for a period
of'time, and when danger threatens, they return to her mouth. After a
few weeks the fry form a small shoal but remain in shallow water
protected by aquatic vegetation from large predators (Jubb, pers. comm.).
The dominant predators are Hydrogyggs, Clarias, crocodiles, otters, and
birds (Crass, 1964; Piennar, 1968). In Lake Kariba, Africa, fish up to
10 cm in total length prefer water depths up to 30 cm on a gently sloping
shoreline, and fish 11 to 19 cm.usually inhabit water 30 to 60 cm
(Donnelly, 1969). Fish leave the shallow nursery areas for the adult
habitat befere they are sexually mature. This movement occurs at 17 to

19 cm at the age of one year [7547 (Donnelly, 1969).

10

In nature fish mature as early as two or three months when about 80
to 90 mm total length (Chen, 1953; Chimits, 1955). Under crowded
conditions they'mature at an older age when only 60 to 70 mm (Jubb, pers.
comm.; Quertermus, pers. observation). The number of eggs spawned by a
female depends upon her size. An 80-mm fish produces 100 to 150 eggs as
compared to 200 to 250 eggs for one of 110 mm, and large females may
spawn over 1000 eggs (Chen, 1953). Under adequate conditions a female
breeds every 6 to 7 weeks (Crass, 1964), or four times (Jubb, 1967),
during the breeding season.

Studies on the growth rate of'T. mossambica indicate different
rates of growth for different waters. In Malaya, under ideal conditions,
fish attain 15 to 20 cm in six months (Chen, 1953). Under crowded
conditions growth rate is greatly reduced. Males are larger than females
at any given age. For example, at 14 weeks males were 61 g and females
only 24 g (Chimits, 1955). In Africa, large males will attain a weight
of 3 kg (Jubb, pers. comm.).

.I.:mossambica is an omnivore. le Roux (1956) reported that fish
under 2 inches showed a marked preference for zooplankton, but that
phytoplankton was the main food of the population he studied and was
already present to a considerable extent in fish under 2 inches. In the
2 to 2.9-inch fish and 3 to 3.9-inch fish, phytoplankton was the most
abundant food, comprising 47% and 77% of the respective stomach contents.
Munro (1967) examined the stomach contents of 170 T. mossambica (18.5 to
47.3 cm T.L.) from Lake Mollwaine, Rhodesia. Filamentous algae and
diatoms (associated with the filamentous algae) comprised 52.1% of the
food contents, while higher plants contributed 31.0%. The remainder was

composed of dipteran larvae and zooplankton. Jubb (pers. comm.) stated

11

that T. mossambica feed on the less coarse aquatic vegetation, algae,
small crustaceans, aquatic larvae, and worms, and at times large
specimens will take small fish. Large males are also cannibalistic
(Jubb, pers. com.).

2. mossambica is well adapted to handle plant material. The mouth
is large with thick lips, protrusible upper jaw, and teeth on both jaws.
The jaw teeth are adapted for scraping the surface of plants and rocks,
and the pharyngeal teeth are adapted for grinding (Kamal Pasha, 1964).
The intestine is long, 103 cm in a 13 cm fish, (Kamal Pasha, 1964), and
the enzyme system is comparable to that of typical herbivorous species
(Nagase, 1964). _T_. mossambica also feeds by shifting bottom material
with its mouth. Gill rakers are reduced to small knob-like projections

(Kamal Pasha, 1964), which indicates a lack of filter feeding.

Economic £51. Scientific Mg

2. mossambica is of considerable economic and scientific importance.
Several detailed behavioral investigations have been carried out
utilizing this species (Baerends and Baerends-Van Roon, 1950; Seitz, '
1948; Neil, 1964 and 1966; and Ruwet, 1968), and recently physiological
studies have been conducted (Solomon and Allanson, 1968; Clemens and
Inslee, 1968; Allanson et al., 1971; Allanson and Noble, 1964; Nagase,
1964; Potts et al., 1967). Their morphological development has been
described by Chacko and Krishnamurthi (1954) and Panikkar and Tampi
(1954).

T. mossambica is cultured in ponds as a source of protein for
humans throughout Asia and many parts of Africa (Chacko and Krishnamurth,
1954; Swingle, 1960). It is the second (behind the carp) most important

culture fish in rice paddies where the annual yield often exceeds 1,000 kg

12

per hectare (Schuster, 1955). In Alabama ponds, production was 2,945.7
pounds per acre in 111 days with initial stocking of 4, 000 fingerlings
(Swingle, 1960). With heavy fertilization and heavy feeding the annual
production in Thailand ponds was as high as 6 to 8 tons per acre
(Pongsuwana, 1956)! T. mossambica is also an important food source in
the areas of its natural distribution where it is primarily captured by
use of ”thrust baskets" (Tinley, 1964).

Due to the herbivorous nature of 2.. mossambica there was hape that
this species could be used to control aquatic vegetation in southern
areas of the United States. But tests at the Fish Farming Experimental
Station, Stuttgart, Arkansas resulted in overcrowded populations and
unsatisfactory control of weeds (Sills, 1970). In Asia and Africa, the
species has been successfully used to control malaria by keeping waters
free of filamentous algae, a necessary breeding substrate for Anogholes
ludlowi (Hofstede and Botke, 1950). The most unusual use of
2. mossambica is its use as a live bait for commercial skipjack tuna

fishing in Hawaii (Shomura, 1964).

lib-

INK

”I.

~\.-

...-rt

METHODS AND MATERIALS

Species Selection

Tilapia mossambica was selected for this investigation because it
is easily bred and maintained in the laboratory. It becomes sexually
mature at an early age, five to six and one-half months under my rearing
conditions. Also, pilot work suggested that'T.:mossambica could be
conditioned to utilize different habitats. The wide distribution of
this species as a result of introductions, and its use in behavioral and
physiological studies were other considerations involved in its selection.
The specimens used for this study were at least fifth generation
laboratory stock. The stock originated from the American Museum of

Natural History, New York about 1965.

When studying the behavior or ecology of an animal, it is important
to understand that animal's perceptual world or "Merkwelt" (von Uexkull,
1934; Klopfer, 1969). Not all physical, chemical, and biological
components of an environment are used by a fish in selecting a habitat
(Sale, 1969b). Therefore, it was important that I utilize a relevant
habitat component for entrainment. Observation of I, mossambica
indicated to me that the substrate is a very meaningful habitat feature
for this species. For example, 2; mossambica establish and defend
territiries on the substrate, manipulate the substrate in nest digging
and feeding, spawn on the substrate, and spend considerable time resting
on or near the substrate. In order to simplify and control the
experiment as much as possible, only the type of substrate was
manipulated and all other habitat components were held constant or

13

14

eliminated. The different substrates were utilized in this study, fine
yellow sand and multicolored stone (particle size 0.8 to 2.5 cm, I =
1.48 cm greatest dimension). These two substrates, differing bothin
particle size and color, were selected to maximize the possibility of
establishing and detecting preferences. Pilot work indicated that these

two substrates were a good choice.

Habitat Conditioning

Fish were raised as a group in 38-1iter aquaria (50.5 cm x 26 cm x
29 cm). These were designated ”conditioning tanks". Each had a slate,
sand, or stone substrate. The "sand tanks” and stone tanks" had 3 to
4-cm depths of sand or stone. Each was continually filtered and aerated
by a LeBern outside filter filled with charcoal and filter wool. The
only object in the tanks other than the substrate substance was the
filter intake tube.

Tap water run through an activated charcoal filter unit was used in
all tanks. Four grams of'marine salt were added to each gallon of water.
The water temperature was maintained at 25 t 2 C. A photoperiod of 15
hours light and 9 hours dark was used (on at 700 hours and off at 2200
hours). Fluorescent ceiling lights were used. All tanks were covered
with glass tops. Black plastic sheets were used to cover the tank sides
to prevent the fish from seeing into adjacent tanks. Fish were fed
BiOrell flake food twice daily at approximately'900 and 1600 hours.

The mouth brooding behavior of _T. mossambica facilitated the
isolation of broods in the conditioning tanks. A female with young in
her mouth was netted from a breeding tank and placed in the desired

conditioning tank. After the young became free-swimming, the female was

Il‘il

15

removed from the conditioning tank. Eight 75-1iter breeding tanks were
available. One to three adult males and two to four females were held
in each breeding tank. This arrangement of breeding tanks insured a

ready supply of "pregnant" females.

Egpgrimental Desigp

The experiment was designed as a threeaway factorial analysis
(Figure 1) utilizing age at experience, conditioning substrate, and age
at testing as the three factors. Three age-at-experience groups were
used. The first group received experience in a sand or stone
conditioning tank the first 60 days after becoming free-swimming. These
fish were then moved to a slate—bottomed conditioning tank. They
remained there until they were placed in an experimental tank to be
tested for a habitat preference. The second group received experience
with a sand or stone substrate the last 60 days before being placed in
an experimental tank. Before this 60-day period they were held in a
slate.bottomed tank. The third group of fish received continuous
experience, from free-swimming to testing, in a sand or stone tank.

Forty fish were tested in each of the three age-at-experience
groups. Twenty were conditioned to the sand substrate and 20 to the
stone substrate. Ten of each set of 20 fish were tested for a habitat
preference at four months of age, and the other 10 were tested at eight
months of age. In other words, 10 fish were tested in each of the 12
experimental groups. The experimental groups are numbered consecutively.
For example, group one had early experience in sand and.was tested at
four months, and group 12 had continuous experience in stone and was

tested when eight months old. Fouremonth fish were 39 mm to 62 mm

l6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cenditioning Age at (months)
Age at Experience Substrate Testing
Early Experience 4
Sand
First 8
60 4
Stone
Days 8
Late Experience 4
Sand
Last 8
6O 4
Stone
Days 8
4
Continuous Experience Sand
8
4
Day one to Testing Stone
8

 

 

 

Figure 1. Experimental

design.

 

17

(x = 50.8) and eight-month fish were 62 mm to 102 mm (x = 72.2) total

length. Males were larger than females.

Eagerimental Testing Tank
Three 230-1iter (120 cm x 56 cm x 34 cm) testing tanks were used to

determine the habitat preferences of experimental fish (Figures 2 and 3).
Each was divided into four equal compartments with opaque, non-reflective,
green plexiglass barriers. The barriers extended from the top to the
bottom of the tank. Each compartment was roughly the size of a
conditioning tank. The ‘walls were lined on the inside surface with non-
reflective, black plastic sheets to prevent individuals from seeing out
or from seeing their own images.

A small doorway (11 cm x 11 cm) was cut at the intersection of the
four compartments in the center of each tank. Each doorway was 18 cm
A above the substrate. Consequently, fish had to swim over a barrier to
pass from one compartment to another. The tanks were filled with water
to a level 2 to 3 cm above the level of the bottom edge of the doorway.
The depth of water used depended on the size of the fish to be placed in
the tank. The shallow water in the doorway was generally sufficient to
prevent fish from lying in the doorway.

Two compartments in each tank had a stone substrate (2 cm deep) and
two had sand (2 cm deep). Diagonal compartments had the same substrate.
The use of two sand and two stone compartments allowed a fish to change
compartments without forcing a change of'habitat. Each compartment had
a LeBern outside filter filled with charcoal and filter wool. The
filters were turned off and their tubes removed when a fish was present.

The water was of the same source as that used in the conditioning tanks.

18

 

Figure 2. Bank of three experimental tanks.

 

 

 

 

Figure 3. Top view of an experimental tank showing location
of doorway and different habitats.

19

‘ A wooden frame was built around each experimental tank and blue
denim cloth was placed on the top, back, and two sides of the frame. A
removable denim curtain was placed across the bottom half on the front
side. This enclosure reduced outside disturbance and the amount of dust
settling into the tanks. A single 40awatt fluorescent tube was suspended
76 cm above the water surface of each tank. The tube extended the
length of a tank and was centered (front to back) overhead to prevent
shadows being cast by the partitions.

Mirrors were suspended at a 45° angle above the tanks for use in
observing the fish. A single mirror, extending the length of a tank,
was hung above the back wall and a small mirror was hung on each side.
This arrangement permitted observation of all portions of a tank when
viewed from a level below the front curtain. The fish could not
directly see the observer during an observation session, and none
appeared to respond to reflections in the mirrors.

The experimental tanks had the same photoperiod as the conditioning
tanks. Their water temperature varied from 21 to 27 C during the course
of the investigation. The fish were fed at approximately 900 and 1600
hours'oy dropping a small portion of BiOrell at the center doorway.

The flakes spread evenly into all four compartments. If not eaten, the

food remained on the surface for several hours before sinking.

Procedures‘§g£_Testing

Fish were tested individually in the experimental tanks. Each was
placed in an experimental tank at about 1600 hours on the day before the
first observation. Each fish was placed in a 600 m1 glass beaker, and
the beaker was then placed in the central doorway. This condition was

maintained fer 30 minutes to permit temperature equalization between the

20

water in the beaker and that of the experimental tank. During this time
the fish could observe all four compartments. The fish was then
randomly'dumped into a compartment and, with the aid of a net, was
shuttled through the doorway three times. This involved a visit both to
a sand and a stone habitat and also provided experience of swimming
through the doorway. Pilot work showed the compartment a fish started
in had no apparent influence on the fish's subsequent selection. But as
a control, in each experimental group, five fish were started in a sand
habitat and five in a stone habitat.

Each fish was observed for a total of 60 minutes over a two-day
period. The first observation was made approximately 21 hours after
introduction (day one). At this time a fish was observed for 15 minutes.
A second 15-minute observation was made 30 minutes after completion of
the first. On day two, after approximately 45 hours of acclimation, two
additional 15-minute observations were made. The procedure was the same
as on day one. The first observation on day one and day two was made
three to four hours after the morning feeding.

The following procedure was carried out on each experimental tank
between consecutive use of fish. Uneaten food was removed. The filters
were run for one hour. The water was thoroughly mixed between ’
compartments. The sand surface was leveled and the stone was rolled.
water was added to maintain the proper depth.

During the observation sessions several variables were recorded:
total time in each habitat, number of visits to each habitat, and
frequency of behavioral measures. Total time in each habitat was
measured with a bank of Gra Lab timers. A visit was defined as swimming
into a compartment after passing through the doorway. A visit was also

scored when a fish swam up into the doorway and then returned to the

21

same compartment. The compartment occupied by a fish at the beginning
of an observation was also scored as a visit. Total time and number of
visits to each habitat are considered as measures of habitat selection.
One or both of these variables were used in the investigations of Klopfer
(1963 and 1965), Wecker (1963), Sale (1968, 1969b, and 1971), and Wiens
(1970).

The following specific behavior patterns were recorded by hand on
a data sheet: swim bouts, total swim.time, ”pivots”, substrate
manipulations, and air gulping bouts. These were recorded separately
for each habitat. A swim bout was considered to be a period of movement.
A period of rest (no movement) of five seconds was used as a criterion
to separate adjacent swim bouts. If the interval between two swimming
periods was less than five seconds, they were scored as one bout. Total
swimming time in each habitat was measured with a stop watch. A "pivot"
is a 600 or greater change in position with no forward movement.
Substrate manipulations entail picking up a mouthful of sand and
spitting it out or mouth-scraping a stone. Air gulping occurs when a
fish swims to the surface and mouths air. This was recorded in bouts.
Five seconds was used as a criterion to separate adjacent bouts. The
above behavior patterns were measured in an attempt to correlate behavior
and habitat selection to test Sale's (1969a) model of habitat selection.

In addition to the preceding variables several additional
observations were recordedeor each fish on each day in the experimental
tank. The color pattern was recorded, because Neil (1964) fully
described the relation of coloration and behavior in T. mossambica, and
color pattern information might be helpful in interpreting my results.

During the observation of the first group (group 11), I noticed that

22

some individuals had not eaten. Subsequently, I recorded whether or not
a fish had eaten the morning food at the time of the first observation
of the day. Mouth-digging in the sand substrate was indicated by the
presence of small pit marks on the sand surface. If the sand surface
was pitted, I knew the fish had been in those compartments, therefore I

recorded the presence or absence of pit marks in both sand compartments.

Statistical Analysis

Three-factor replicated analyses of variance were run on time in
the sand habitat and time in the conditioned habitat (Sokal and Rohlf,
1969). Although these two analyses (i.e. time in sand habitat and time
in conditioned habitat) are redundant in that they utilize the same
values or the complement of the values, they are both necessary to
clarify the results. The data fulfilled the assumptions of the analysis
of variance without transformation. The analyses were done using time in
seconds, but the results are presented in percent time on the tables to
facilitate making comparisons between groups. A priori individual degree
of freedom comparisons were run on the experience treatment utilizing
time in the conditioned habitat. The analysis of visits to different
habitats was done by chi-square independence tests. The nature of the
behavior results made statistical analysis of behavioral comparisons

unnecessary.

RESULTS

Effggt'gf Conditioning

No sex differences were found for eight-month fish; so data for
males and females were combined (e.g. for groups 2, 4, 10, and 12, males
spent 65.6% and females spent 62.5% of the time in the conditioned
habitat). Sex determination of the four-month (immature) fish was not
attempted since no sex differences were found in the data for eight-
month fish.

Table 1 gives the average percent time spent in the sand habitat.
To determine the effect of conditioning, comparison should be made
between fish of the same age and age at experience but conditioned to
different habitats (e.g. groups one and three or two and four). A
difference of zero or near zero for a group comparison indicates no
effect of conditioning. On day one there was no effect of conditioning
for early or continuous-experience groups. For the late-experience fish
a positive effect of conditioning is suggested for four-month fish on
day one, but for eight-month fish the difference does not show a
positive conditioning effect. Naive fish (i.e. no prior experience with
sand or stone) spent 53.3% of the time in the sand habitat on day one
(Nsao fish).

On day two the effect of prior experience on habitat selection is
apparent. All group comparisons show positive conditioning for the
earlyaexperience and continuous-experience groups. The sand-conditioned
groups (1, 9, and 10) are close to random (50%) suggesting that only the
stone-conditioned fish were positively conditioned. However, tests of
naive fish resulted in their spending only 30% of the time in a sand
habitat on day two (Ne30 fish); therefore, a value of 50% for sand

23

24

.Aosp hub no bow poop omemv m cap but one acre hep
.oco hob no oceans: been cop ca pzomm asap pzooaom owcno><

.H canoe

25

.039 hoe one one hot 20 been we seem mo gooaw

seem Bonn poz*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NH no.“ can a com w mama
96$ 8
HA and: 0.8. can 3
ea see one one n so as as
n
825%
a a. as m . me a . an a 88528
n if t: can n new
mzoem so
a com «.9 oém a
b New ES cém m SS
92%
. . . Bzfimmfie
m c on o me c on a Bag
a ass can can a mean
seam
m exam mam fee a om
N «.8 0.3 0.3 m 9mm:
.- I 95
H 9% firm fun a 8%ng
Ema
a 0% :5 0% was as use
a S ofiamfi 93.8% 825%
Some .2 8a. eszoEHmzoe .2 Be.

 

BdHHmdm nadm 2H mZHe % mu<mm><

 

 

 

 

26

conditioned fish on day two does not mean there was no positive
conditioning. Conversely, the results from the naive fish suggest that
a value below 30% would be needed for stone-conditioned groups before
positive conditioning can be assumed.

The results for the late-experience groups differ from the early
and continuous—experience groups. The four-month group comparison
indicates no effect of prior experience. The eight-month groups show a
substantial effect of conditioning, but the effect is negative in that
these fish avoided the conditioned habitat. This avoidance was seen for
both the sand and stone-conditioned fish. This reversal was unexpected
and difficult to believe; so an additional 10 fish were tested in each
group. Again the rejection of the conditioned habitat was present. In
this case the observed difference between the sand and stone-conditioned
groups for time spent in the sand habitat was 32.5% as compared to 51.8%
for the original group comparison. The 19.3% difference between the
replicates may be due to the fact that the second replicate was done
with loqmonth fish with four months of late experience.

The analysis of variance (Table 2) for time spent in the sand
habitat indicates the major component of variance is in the second-order
interaction. Figure 4 is a graphic presentation of this interaction.
The graphs for early experience and continuous experience show the
positive effect of conditioning in these groups (sloped lines). A
strong interaction is seen on the late-experience graph. The horizontal
line for the fouremonth fish reflects the lack of conditioning effect.
The crossing of the lines (at a 45° angle) is due to the rejection of

the conditioned habitat by the eight-month fish.

27

Table 2. Analysis of variance table for time spent in the sand habitat

on day two.

 

m
A (Experience)
B (Habitat)

C (Age)

A x B

A x C

B x C

A x B x C
Within

Total

g;

2

F4 FJ

108

119

§§
1,077,489
1,182,862
312,120
6,698,430
303,546
1,064,460
72,397,858
70,898,622

.Mi .E
538,744t5 0.82
1,182,862.0 1.80
312,120.0 0.48
3.349.215.0 5.10
151.7710 0.23
1,060,460.0 1.62
36,198,929.0 55.14
656,468.?

n.s.
.005<p<.01
n.s.
n.s.

p<.001

 

5.33%: been 93 2H. 08.3. no.“ 90302093 noononooooom Ho ooflmozomcnm 9.2de .e chomfim

ocean econ

 

 

 

oucotooxm escoczcou

h<tn<z OmZOCEZOU

one: once

 

oucctodxm 20..

 

UCO~m

econ

 

1

 

eucotodxm beam

 

OQNOWVGN"

(801 x ass) lvusvw <1st NI awu '3/w

29

Day two F (i.e. fish that fifed‘ on day two) for the early and
continuous-experience groups generally shows a greater positive
conditioning effect than that of day two (Table 1). Also, the four
month late experience groups show a positive effect of conditioning on
day two F, which was not present on day two. The rejection of the
conditioned habitat was still strong on day two F for the eight-month
late-experience fish. The four-month group with continuous experience
in a stone habitat is not the same group as used on day one and day two.
This substitute group for day two F was hyperactive compared to other
groups, and had a severe fungal infection and subsequent medication
during rearing. Therefore, less confidence is placed in this group.
None of the other groups experienced any apparent disease during their
rearing. Seven eight-month naive fish spent 100% of the time on day two
F in a stone habitat.

Table 3 gives the total number of visits to a sand and stone habitat
for all fish in each group. These results were analyzed by chi-square
twoaway independence tests and are summarized in Table 4 (part II).
There were no age differences for early and continuous-experience groups
(Table 4, part I); so four and eight-month groups were combined in
part II. However, there was an age difference for late-experience fish
so the analysis was run separately for four and eight-month late-
experience fish.

There is no significant effect of prior experience for any group
comparison on day one. Earlyhexperience and continuous-experience group
comparisons show a significant effect of prior experience on day two.
This positive effect is also evident in Table 3. Here seven of eight

treatment groups show more visits to the conditioned habitat, and 63.4%

30

.Aosp moo so oom poop omfimv m 03p hco one .osp hmo
.coo hco so mpcpfinco ozone one once op mpfimfi> Mo nonsoz

.m oaflme

31

.ozp huo one one moo no oomn we seem mo anonm cede scam 902*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OH N 3H m OH 0H m uZHBmma
. . . .. . .. - . E03
09
*wn *HN 0H # 0H 0H d
mzo Had
A m NH OH mH OH m
Qz¢m EUZMHmmmNm
3 3 NH 9 a: S s 885928
m NH m “H MH m m mwda
mzoem
as a be 9 cm 3 a on
Bmdfl
H m B” m 3H m w
Qzem muzmHmmem
a a a e S on a an:
HH m MH h NH NH m mwdn
mZOBm
an n on m ea 3 a on
BmflHh
N 0H m MH NH m m
024m muzmHmmmNm
: HH m 3H 0H NH 3 Mum<m
rammed on mmdm op emcee op owned op omopm on once on
speed» mpwma> mpamfie mpflmfi> cpfimfi> muameb GZHBmmH BHHHmHm MUEMHMMKRW
a ea. .3 ea. ea .8 ea .2 8a. ufizogeso .2. 8a.

 

 

32

Table 4. Analysis of habitat visits by chi—square two-way independence

tests.

 

I.

II.

III.

4rmonth vs. 8 month fish comparing visits to the conditioned and
unconditioned habitat

A. Early Experience (sand and stone conditioned fish combined)

 

 

chiasquare probability
Day one (n=40 fish) '0.103 n.s.
Day two (n=40) 0.333 n.s.
Day two F (n=30) 0.045 n.s.
B. Late Experience (sand and stone conditioned fish combined)
Day one (n=40) 2.253 n.s.
Day two (n=40) 5.467 .025-7p7.01
Day two F (n=28) 6.533 .025>p7.01
C. Continuous Experience (sand and stone conditioned fish combined)
Day one (n=40) 0.386 n.s.
Day two (n=40) 0.685 n.s.
Day two F (n=30) 2.513 n.s.

Sand conditioned vs. stone conditioned fish comparing visits to the
sand and stone habitats to show effect of conditioning

A. Early Experience (4 month and 8 month fish combined)

Day one (n=40) 0.104 n.s.
Day two (n=40) 9.670 p<.005
Day two F (n=30) 10.510 p<.005

B. Late Experience
4 month fish:

Day one (n=20) 0.997 n.s.
Day two (n=20) 0.000 n.s.
Day two F (n=13) 0.972 n.s.
8 month fish:
Day one (n=20) 0.968 n.s.
Day two (n=20) 9.166 p<.005 (reversal)
Day two F (n=15) 5.704 ‘.025>p>.01 (reversal)
C. Continuous Experience (4 month +-8 month fish)
Day one (n=40) .000 n.s.
Day two (n=40) 4.889 .05>p> .025
Day two F (n=30) 6.095 .025>p) .01

Early experience vs. continuous experience comparing visits to
conditioned and unconditioned habitat to show differences in
experience treatment

A. 4 month fish (sand + stone conditioned fish)

Day one (n=20) 0.038 n.s.

Day two (n=20) 0.000 n.s.

Day two F (n=13) 0.571 n.s.
B. 8 month fish (sand + stone conditioned fish)

Day one (n=20) 0.011 n.s.

Day two (n=20) 1.461 n.s.

Day two F (n=15) 0.461 n.s.

33

Table 4. (cont'd)

4
r

IV. Early experience vs. late experience comparing visits to conditioned
and unconditioned habitats to show differences in experience treatment

A. 4 month fish (sand + stone conditioned fish)

Day one (n=20) 0.730 n.s.

Day two (n=20) 1.610 n.s.

Day two F (n=13) 0.450 n.s.
B. 8 month fish (sand + stone conditioned fish)

Day one (n=20) 0.021 n.s.

Day two (n=20) 15.310 p<.005

Day two F (n=15) 10.520 p< .005

34

of all visits for these 80 fish were to the conditioned habitat. The
chi-square values for the early and continuous-experience groups are
greater on day two F than day two. This suggests a greater positive
expression of conditioning on day two-I? than on day two. Furthermore,
on day two F (Table 3) all eight groups show more visits to the
conditioned habitat, and 66.1% of all visits were made to the

conditioned habitat by the 60 fish in the early and continuous-experience
groups.

The four-month late-experience groups indicate a slight positive
effect of conditioning on day two F (Table 3) but it is not significant.
The eight-month lateaexperience fish show a highly significant (p<(.005)
but negative effect of conditioning on day two. Day two F is also
significant (.025 >p7.01) but the chi-square value is much lower
suggesting that the reversal was not as strong for fish that fed.

Naive fish gave the following results for the number of visits to
sand and stone habitats. On day one, 30 of 65 visits were to sand
(N:30 fish). On day two, only 15 of 60 visits were to sand (N?30 fish),

and on day two F zero of 14 visits were to sand (N=7).

m 2f Age 2.; Experience

Table 5 gives the average percent time in the conditioned habitat
for day one, day two, and day two F. The analysis of variance was run
on total time on day two and the results are presented in Table 6. The
greatest source of variance is in the second-order interaction, which is
caused by the reversal of the late-experience eight-month groups. A
graph of this interaction would be similar to Figure 4 for the analysis

of time in the sand habitat.

35

.Aozp heo no ooh peso anMv m can heo one .osp heo .ono
heo no pepfloen oonoflpfionoo nfi anode asap pnoonom oweno><

.m oHQeB

36

 

63...? heo one one heo no omen ee neg .«o macaw caee non.“ $2,:n

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NH new can can m cfiamfi
Sean
.2 seem 0.8 can a 8
. as be
3 line was in: m ,
ozen 823%
-m fine new lime m use
mean
a 11.2 new 98 a on
$3
w New ~23 eém m
92m 823%
m 0.0m 0.? com a 83
a dam ode to: m mica
alone
m sea sea. is a on
Home
N mam of. 0.3 m
ezsn BEBE
H new Hem new a Sea
a, a can in can on mac no seamen gamer BEE .
Bee .2 Be $2033on 3 8e.

 

 

 

 

names @033on am one a tense.

 

 

 

 

Table 6. Analysis of variance table for time spent in the

37

habitat on day two.

conditioned

 

some. if.
A (Experience) 2
B (Habitat) l
C (Age) 1
A x B 2
A x C 2
B x C 1
A x B x C 2
Within 108

Total . 119

§§_ MS

6,698,430 3,349,215.0 5.10
1,930,403 1,930,403.3 2.94
1,064,460 1,064,460.0 1.62
1,077,489 538,744.5 0.82

In:

1,499,237 749.6186 1.14

312,120 312,120.0 0.47

71,202,168 35,601,083.9 54.23
70,898,622 656,468.?

.005 (p (.01

 

38

The age-at-experience treatment source of variance is significant
on day two (Table 6). Individual degree of freedom comparisons were
carried out on the experience treatment (Table 7). There is no
significant difference for early as compared to continuous experience.
But there is a highly significant (p‘<.001) difference between early or
continuous experience as compared to late experience. The late-
experience group showed no effect of conditioning for four-month fish
and a negative effect of conditioning for the eight-month fish
(Table 5).

Table 8 summarizes the time variable for all groups that showed
positive conditioning on day two (groups 1-4 and 9-12). All of these
eight groups had experience with the sand or stone habitat during the
first 60 days of life. On day one the average percent time in the
conditioned habitat (48.5%) and the percent of fish (49.3%) that spent
more time in the conditioned, as Opposed to the unconditioned, habitat
were very close to random (50.0%). On day two these two figures (64.5%
and 68.8%) reflect the positive effect of conditioning. The increased
positive effect on day two F is also apparent (67.1% and 74.1%). The
figures for day two F include the substitute group 11 fish. If these
were removed from the analysis the difference between day two and day
two F would be greater.

Table 4 (parts III and IV) summarizes the statistical analysis for
habitat visits in order to show the effect of age at experience. Part III
indicates no significant difference between early and continuous
experience for both four and eightamonth fish on any day. The comparison
of early versus late experience (part IV) indicates no significant
difference for fouremonth fish but a highly significant difference
(p‘<.005) for eightamonth fish on day two and day two F.

39

Table 7. A priori individual degree of freedom comparisons for
experience treatment (factor A) for day two time in
conditioned habitat.

 

 

Source of Variation Q §§ _M§ E
Experience 2 6,698,430 3,349,215 5.10 .005<p<.01

Early vs. Continuous 1 369,377 369,377 0.56 n.s.

Ear1y+Continuous 1 6,329,053 6,329,053 9.64 p<.001
vs. late .

Error 108 70,898,622 656,469

 

40

Table 8. Summary of time variable for all groups with sand or stone
experience during the first two months (groups 1-4 + 9-12).

 

AVERAGE % TIME IN % OF FISH THAT SPENT MORE
DAY CONDITIONED HABITAT TIME IN CONDITIONED HABITAT
IHAMUNCONDITIONED HABITAT

 

 

 

DAY ONE 48.5 49.3
(N=80 fish)
DAY Two 64.5 68.8
(N=80 fish)
DAY 'IWO F 67.1 74.1

(N=560 fish)

 

 

 

 

IKD

41

Behavior Analysis

In order to clarify the lack of conditioning effect on habitat
selection on day one as compared with day two, I compared the activities
on these two days. Table 9 indicates no definite differences in
swimming time, number of swim bouts, or number of pivots on day one and
two, particularly for groups 1-4 and 9-12, the groups that showed"
positive conditioning.

Table 10 compares the behavior of fish in the conditioned and
unconditioned habitats. The activity levels were corrected for unequal
time spent in the two habitats by dividing by the percent time in the
corresponding habitat. The eight groups that showed a positive effect
of conditioning (groups 1-4 and 9-12) were used in this analysis. ‘When
total activities were used, there was a slightly higher activity score
for swim.bouts, pivots, and substrate manipulations for fish in the
familiar (conditioned) habitat as opposed to the unfamiliar habitat.

The amount of time swimming was greater in the unfamiliar habitat.
However, if the total score for each group is used there was no apparent
difference in activity for fish in the familiar and unfamiliar habitat.
Thus, an equal number of groups had a higher level of activity in the
unfamiliar and familiar habitat.

An analysis was done to determine possible differences in behavior
frequencies given by fish while in the sand and stone habitat (Table 11).
Frequencies were again corrected for time in each habitat. Swimming
time and number of swim bouts were vfgrea'ter‘a- in the sand habitat, but in
both cases only 7 of 12 groups had greater frequency scores in the sand
habitat. The number of pivots per unit time was slightly greater in the

sand habitat, but only 5 of 12 groups had.more pivots in the sand.

42

Table 9. Activity comparison of fish on day one and day two. Numbers
in parentheses indicate number of fish more active on the

indicated day than on the alternate day.

 

~ TOTAL SWIM TOTAL SWIM TOTAL PIVOTS
TIME (SECONDS) BOUTS

 

DAY ONE DAY TWO DAY ONE DAY TWO DAY ONE DAY TWO

 

ALL GROUPS 7450 5908 844 848 1195 1063
(45) (43) (44) (36) (58) (49)

 

GROUPS 3692 4023 505 580 811 721
1-4 + 9-12 (28) (30) (27) (27) (36) (38)

 

 

 

 

 

 

43

 

 

 

 

 

 

EH3 minors so an

 

mHH3 mmbomw ho *

 

HBH>H90< Adyoa

 

MEH>HBU<.A¢HOB

 

m n and 8.: zoflfionnfin

Samoan

m m 3.3 3.2” Sofia

a a 3.2 RAH 88m

Em .

e e 31% 3.1.3 mam.

Em
Samoa page 2593 93.8%
$32.35 an one See manque: 2H E 9st mfiqgaze. THEE

moi BEBE. annihilatew and @953 magma an and. a u E was a m

 

Acne. heo one one heov NHIm one dIH mmnonm no.“
npeflhfi neflHBemno one neHHHnem n.“ now.“ he noewaeneoc nogeoom .3“ oHoee

 

 

 

 

 

 

 

 

 

 

 

Aonou we moan NV
N N SN 36 mzoflgfizea
Samoan
N m RAN NAN Bend
m N no.3 2.8 Room firm
m N i 85H 363 $9588
95 fizm
SS3 Samar
macaw an Bee 5
E SE moi Nana ES and 2.93% seamen
Emacs awesome Emacs. awesome ozone an ease fi
EH3 Booms EH3 $85 one N a one a a
so $952 so one; Seneca .58 Edged .32. —

 

.Aedoonw HHev epepfioen onopo one onee onu na
ooeoonnxo enowoeoon non noeflnedeoc hononoonm noe>enom .HH oHofie

 

45

Substrate manipulations were considerably'more frequent in the sand '
habitat as measured by total activity and number of groups. Eight
groups did more substrate manipulating in sand and only two did more in
stone. Two groups tied at zero.

The preceding analysis suggests that there may be a quantitative
difference in behavior patterns observed for fish conditioned to the
two different habitats. Table 12 compares the total scores for fish
conditioned to the sand and stone habitats. These comparisons were made
using all groups for both day one and day two. Stone-conditioned fish
spent more time swimming, had a greater number of swim bouts, and did
more substrate manipulating. The substantially higher scores for stone
conditioned, compared to sand conditioned, fish for these activities is
clouded by the fact that only four of the six group comparisons indicate
a greater activity for the stone conditioned fish. For all three
patterns it is the continuous-experience groups that do not coincide
with the early and late-experience groups. Pivoting was the only
behavior expressed at a greater frequency'by sand-conditioned fish, but
only three of six group comparisons show a greater pivoting rate for
sand-conditioned fish.

During the course of the observations it was obvious that four-
month fish were more active than eight-month fish. An analysis was
subsequently'done to verify this observation (Table 13). Four-month
fish did twice as much swimming and had almost twice as many swim bouts
as compared to eightemonth fish. Substrate manipulations were also more
common for fouramonth fish (360 compared to 85). The number of pivots,

however, was almost identical for four and eight-month fish.

 

 

 

 

 

 

 

 

Aoommoooo monaoam
an Boon 2H mono
o No a mm NN BEE ma ..8 $985
ozoflesfizaz
BEBE
o no N Sod BNH macaw
o no a , 33 SN Boom Eon
o No a mNNm neon 8mg some seem
moose 8289328 mzoon on Ban on
ozone an em . @658on 8289328
Home: madame Ens mono em BE am
ozomHmseaoo Bore Seneca .38 BEES. .298
8282828 Boom
.m> omzoEBzoo
ozem no mmesz

 

 

 

 

.oonwnaoo 25. heo one one heo no
mason» fine now new.“ oonogfionoc oncoae one onee mo noownenaoc nogenom .NH oHnem.

 

Table 13.

4?

Behavior frequency comparison of four-month and eight-month

fish for all experimental groups combined.

 

# of 4 vs. 8
MONTH COMPARISONS WITH GREATER

 

 

 

 

 

 

 

 

 

4 MONTH 8 MONTH .ACTIVITY IN 4’MONTH GROUP
SWIM
TIME (SEC) 8936 4422 4 of 6
SWIM
BOUTS 1108 674 5 of 6
PIVOTS 1108 1151 2 of 6
SUBSTRATE 360 85 4 of 6
MANIPUIATIONS (44)* (19)* (5 of 6)*
SWIM BOUTS 1.00 0.59 5 of 6 4 vs. 8 month
: PIVOTS ' ' Comparisons show”4 month with
' " relatively more swim.bouts.*

than pivOts. »

 

 

* Number of 15 minute observation periods in which the behavior

was expressed.

9212; Pattern Analysis

The color pattern of each individual was recorded daily at the time
of each observation. I analyzed these patterns on the basis of whether
or not they occurred in the conditioned or unconditioned habitat
(Table 14). A hatched or striped pattern was less common than neutral,
barred, or dark for fish in both the conditioned and unconditioned
habitat on day one, day two, or day two F. But a hatched or striped
pattern was relatively'more frequent for fish in an unconditioned, as
Opposed to those in a conditioned, habitat. This difference is not
significant for any'day, but approaches significance on day two F
(0.1 >x2> .05). On day two F, 38% of the fish in the unconditioned
habitat were hatched or striped whereas only 16% were hatched or striped
in the conditioned habitat.

The ability of rapid color change by _T_. mossambica was evident ‘
during this investigation. Individuals Changed oblor intensity within

one minute after changing habitats, with fish being much lighter in

color in a sand habitat than a stone habitat. However, I did not observe

' a color pattern Change (e.g. barred to hatched) with a change of habitats,

although'T, mossambica is also capable of a rapid color pattern change

(Neil, 1964 and Quertermus, personal observation).

 

 

Table 14. Comparison of coloration of fish selecting conditioned and
unconditioned habitats for groups 1-4 and 9-12.

 

 

 

 

 

 

 

 

 

 

{
In Conditioned Habitat In Unconditioned Habitat
Number of Number of Number of Number of
Fish Fish Fish Fish
Hatched Neutral, Hatched Neutral,
of Barred, or Barred,
Striped or Dark Striped or Dark
Day
One 2 32 7 27
Day
Two 16 34 8 14
Day

 

 

 

DISCUSSION

Igportance‘gf,Early Egpgrience
The results of this investigation indicate that prior habitat

experience may be of primary importance in establishing a preference for
a resting habitat in certain fish. Under the conditions of this study,
prior habitat experience played a significant role in the habitat
selection of‘T. mossambica. However, not all regimes of experience were
sufficient to establish a habitat preference as judged on the basis of
amount of time or number'of visits to a habitat. Fish given experience
only late in life (groups five through eight combined) did not select
the conditioned habitat, whereas fish given only early experience (first
60 days) or those that received continuous exposure to a habitat
selected the conditioned habitat. In other words, early experience was
not only a sufficient but also a necessary condition for habitat
selection.

The four-month fish given experience late in life (groups five and
seven) were positively conditioned to a habitat based on visits for both
day two and day two F, but no influence of experience was seen when
determined by the time variable except on day two F. This weak positive
effect of conditioning (compared to groups 1-4 and 9-12) may have
resulted from the fact that although these fish received late experience,
this still occurred before they became sexually mature. In other words,
for‘T, mossambica, habitat fixation seems to occur before individuals
'become sexually'mature, and the strongest attachment to a habitat is made
during the first few weeks of life.

These results seem logical.when one considers what is known of the
natural history of this species (see Natural History Section). Females

5,0

51

carrying young move to shallow water areas with vegetative cover. Here
the young are released and are then guarded by the female for a period
of time. Fixation to a habitat may well occur during this period. .After
the young are abandoned by the female, they would remain in this habitat
with relative safety from predation. As the young grow, the number of
potential predators is reduced allowing movement from the areas of cover.
Then, if danger threatens, the response would be to return to a place of
safety which was learned earlier in life. There may be other advantages
for habitat fixation besides possible avoidance of predation. For
example, fixation to a prOper habitat may be related to the presence of
an adequate food supply. These hypotheses should be tested with
detailed field work. In nature, the vegetative component of the habitat
may be important in habitat fixation and should be examined.

The present study'is one of the first to show that early'experience
is not only a sufficient but also a necessary condition for determining
later behavior. Mbst investigations of early experience and subsequent
behavioral responses have been poorly controlled. The usual procedure
has been to give an animal some type of early experience, test at a
later time, and then conclude that early experience is important. The
control is to give the same experience at other times of life to
determine whether the same effect is noted (see King, 1958 for further
clarification). Mbst of the "early experience" studies investigating
habitat selection in vertebrates suffer from this lack of control
I (e.g. Klopfer, 1963 and wecker, 1963). A notable exception is Wien's
(1970) work with tadpoles. He concluded that early experience was
sufficient to establish a habitat preference but was not a necessary

condition, since tadpoles given the same length of experience later in

52

life gave the same response. Thus, my study is the first to show that
early experience is a necessary condition for the establishment of a
preference for a resting habitat in a vertebrate.

It is of interest that continuous experience did not cause a
stronger fixation to a habitat than early experience. In fact,
observations of early-experience fish suggested a greater (but not
significant) effect of conditioning than for continuous-experience fish.
Earlyaexperience fish averaged 68.3% and continuous-experience fish
averaged 60.8% of the time in the conditioned habitat on day two, but
the difference between early and continuous experience was not as great
on day two F. Also, based on visits there was even less difference on
day two between early and continuous-experience groups (i.e. 66.7% and
62. 5% of the respective visits were to the conditioned habitat). The
rfact that continuous experience did not increase habitat fixation over
early experience is another indication that only early experience (first
two months or less) is important in establishing a habitat preference.

The avoidance of the conditioned habitat by the eightemonth late-
experience fish is puzzling. The eightamonth continuous-experience fish
also had immediate prior experience in the conditioned substrate, but
they did not reject it. However, the somewhat lower scores for the
continuous-experience fish compared to the earlysexperience fish may be
due to a slight avoidance of the conditioned habitat by some individuals.

Two possible explanations may account for the avoidance by the
late-experience fish. Since these individuals received early experience
on a slate substrate, the avoidance for the habitat experienced late in
life may'be due to a specific search.image (Tinbergen, 1960 and Hinde,

1966) for the ”earlyhconditioned slate habitat”.

53

The second possibility is that the exploration of the novel
(unconditioned) habitat was not as complete by the end of the second day
for the late-experience fish as it was for the early and continuous-
experience fish. Berlyne (1966) and others (see review by Fowler, 1965)
have shown that the amount of exploratory'behavior shown by an animal
(in this case a laboratory rat) depends on the novelty of the test
environment. The more novel the environment the more intense and longer
the exploration. Perhaps the late-experience fish found the unfamiliar
habitat more novel than did early or continuous-experience fish.

Although the avoidance of the conditioned habitat is real on day
two, it may not persist indefinitely. Preliminary results of a follows
up study suggest that the avoidance disappears by the fifth or sixth day
in the experimental tank, when fish begin to select the conditioned
habitat. Ten individuals with late experience on stone spent 65.0% of
the time in a stone habitat on days five and six as opposed to only 28.5%
on day two. These results are questionable, since seven of the 10 fish
were used earlier in the experimental tanks and all 10 were 10 to 11
months old with four to five months experience on stone. This study is
continuing with another group of fish. If these results are verified,
the second explanation for the avoidance seems the most likely.

Naive fish (no experience with sand or stone), tested for a
selection of a sand or stone preference, strongly selected the stone
habitat (70.0% on day two, N =‘30 feur and eight-month fish, and 100% on
day two F, N = 7 eightamonth fish). There are several possible
explanations for these results. 'T.:mossambica may have an ”innate"
preference for a habitat that more closely resembles the more

heterogeneous stone substrate. Another possibility is that the slate-

51+

bottomed tanks in which these fish were raised more closely resembled the
stone substrate as perceived by the fish. A third explanation relates

to the more heterogeneous nature of the stone as compared to the sand
substrate. Animals generally explore more vigorously, and for a greater
length of time, objects that offer more varied or irregular stimulation

(Berlyne, 1966) .

Comparison _q_i_‘_ 92E 9_n_e_, _Tw__g, 313 E2 2:

The results for day one, day two, and day two F were markedly
different (Tables 1, 3, and 5). There was no indication of positive
conditioning on day one (except group five). There are two possible
reasons for the random distribution of fish on day one. The fish may
still have been :‘in an exploratory phase on day one, but by day two
exploration of the available environment was complete and a selection
was finally made. This hypothesis is weakened by the lack of a
significant difference in activity levels on day one and day two
(Table 9). If an exploratory phase prevailed on day one (22 to 21+ hours
acclimation) and not on day two, there. should have been greater
activity on day one compared to day two when a selection of a habitat
was made.

An alternate hypothesis is that the fish were "stressed" on day one
and consequently failed to discriminate or failed to select a habitat.
Recent studies have shown the simple capture and handling of fish is
sufficient to cause an increase .in blood glucose (Chavin and Young,
1970) and changes in plasma protein concentrations, hemoglobin
concentration, and erythrocyte size (Bouck and Ball, 1966). Chavin and
Young (1970) have shown that the simple net transfer of goldfish from
one tank to another produces a significant hyperglycemia of two days

i

55

duration. Such physiological changes, associated with stress, could be
reflected in the behavior of fish. Fish.may also be stressed by factors
other than handling. Stress may result from.being in an unfamiliar
habitat. Consequently, fish remaining in an unconditioned habitat would
have remained stressed for a longer period of time than fish in a
conditioned habitat.

The fact that there was a greater positive effect of conditioning
for fish that fed on day two (day two F) than for all fish on day two
tends to support the stress hypothesis. Those that fed were presumably
the least stressed, and these fish exhibited the strongest habitat

preference.

Activity and C_olo_r Pattern

Sale's (1969a) hypothesized mechanism of habitat selection predicts
that exploration activity should be greater in an inadequate (non-
preferred) habitat while non-exploratory patterns should be greater in
an adequate (preferred) habitat. His study on juvenile manini, a
Hawaiian reef fish, fits these predictions fairly well. To test his
hypothesis with the data from this study the activity scores for fish
while in the familiar (selected) and unfamiliar habitats were compared
(Table 10).

I consider swim time, swim.bouts, and pivots as possible measures
of exploration. Substrate manipulation seems to be related primarily to
feeding and therefore non-exploratory. Based on total activity scores
the swim.time and substrate manipulation results fit Sale's model, but
swim.bouts and pivots do not. However, on a group basis, there was no
difference in activity frequencies for fish in familiar and unfamiliar

habitats. NW'results do not affirm.Sale's hypothesis, but this may be

56

due to several factors. I did not make observations during the first

few hours of acclimation, as he did; the species were different; the
behavioral.measurements were not identical; and in my experiment the fish
had a choice of habitats, whereas in Sale's experiment fish were allowed
access to only one habitat at a time. Sale made his observations only

65 minutes after a fish was placed in a new habitat. It is interesting
that he got significant results with such a short acclimation period.

Regardless of the type of prior experience, the behavior measures
of swim.time, swtm'bouts, and substrate manipulations, based on total
activity (Table 11), were considerably'greater for fish in a sand, than
stone, habitat. However, only for substrate manipulations was there a
distinct difference based on. number of groups. The greater" amount of
swimming in a sand habitat may be due to the less complex visual
appearance of this substrate as compared to the stone. Consequently,
more movement might be necessary for sufficient visual stimulation in a
sand habitat. The greater". amount of substrate manipulation of the sand
was probably a result of the nature of the two substrates, since the sand
could be picked up, mouthed, and spit out, while the stone could only be
scraped.

Stone-conditioned fish swam.more, had more swim.bouts, and did more
substrate manipulating than did sand-conditioned fish (Table 12). This
difference between sand and stone-conditioned fish was true regardless
of the substrate the stone-conditioned fish were visiting. These
results suggest that prior habitat experience can effect subsequent
behavior of fish. Other studies have shown that habitat experiences
influence behavior of’animals. wecker's (1963) data show that groups of

Perggyscus maniculatus given different prior habitat experience had

57

different periods of activity, rate of travel, and depth of penetration
into neighboring habitats. KlOpfer (1967) and Sheppard, Klopfer, and
Oelke (1968) found differences of feeding behavior for island and.main-
land populations of several bird species.

The greater activity by stone-conditioned fish may be a result of
rearing in a more complex habitat, since the stone substrate is more
heterogeneous than sand. Luchins and Forgus (1955) reported that rats
reared during infancy in varied environments exhibited greater activity

in Y mazes when tested at maturity than did restricted rats.

 

Four-month fish were substantially more active than eightemonth
fish (Table 13) on the basis of swim.time, swim bouts, and substrate
manipulations. These results seem reasonable. Greater exploratory
activity of’young (not infant) compared to older animals of a species
has been observed frequently (welker, 1961).

Pivoting was done at the same rate by four and eight-month fish.
As a matter of fact, for all behavioral comparisons (Tables 9, 10, ll,
12, and 13) the pivoting rate was identical or similar. This suggests
that pivoting is an important behavior regardless of age, experience, or
habitat. During the observation sessions, most of the fish spent the
majority of the time resting on or near the substrate in the center of
a compartment (habitat), with the "typical" fish occasionally pivoting
or, less often, swimming for a few seconds. Perhaps pivoting functions
as a monitoring behavior for fish to maintain their "bearings" and
perhaps to keep watch for predators, food, etc. Additional work is
needed to test this hypothesis.

Neil (l96h) found that a hatched color pattern was typical of

extreme "fright" in l. mossambica. The color pattern analysis (Table lb)

58

indicates that a hatched or striped (often difficult to distinguish from
hatched) pattern occurred in a greater (but not significant) percentage
of fish in the unconditioned habitat compared to the conditioned habitat.
The difference between the conditioned and unconditioned habitat
approached significance on day two F. In other words, fish without a
hatched or striped pattern were observed.more frequently in the
conditioned, than the unconditioned, habitat. If the absence of fright
colors patterns is an indication of lack of stress, either unstressed
fish were more likely fix) have selected the conditioned habitat, or fish
that selected correctly changed to an unstressed color pattern. The
latter suggestion is weakened by the fact that I did not observe a change
in color pattern as fish changed habitats during the observation sessions.
The former suggestion is favored by the findings that: (1) fish selected
the conditioned habitat on day two'but not on day one, and (2) fish that
fed on day two showed a stronger selection for the conditioned habitat

than those that did not feed.

2. mossambica _a_s; a Representative Fish t_o Illustrate Habitat Selection.
Was I. mossambica a good choice to demonstrate fixation of a habitat
based on prior experience? There seems to be a basis of argument for
both sides. The young of animals that require parental care have a more
plastic behavior (Klopfer and Hailman, 1965) and presumably are
influenced by experience to a greater extent. Also, fish that remain
for a period of time with their parents or parent are more likely'to
develop responses to environmental stimuli than species with no parental
care. Based on this reasoning T. mossambica was a good choice.
Klopfer (1965) states that species preferences are proportionally

narrower the greater the number of similar species coinhabiting an area.

59

'_I:. mossambica apparently is found with few if any similar sympatric .-
species and should have relatively wide preferences. The fact that

2. mossambica has such a wide distribution and environmental tolerance
suggests that this is true. This point suggests that T. mossambica can
not be as strongly fixed to one type of habitat as a species with narrower
preferences .

The use of a laboratory population for this study may have presented
problems. Price (1970) reported that wild populations of P. maniculatus
were more hesitant to enter unfamiliar areas than were laboratory
populations. Therefore, the use of a field population might have yielded
stronger positive results. However, Wecker (1963) found that laboratory
stocks of _P. maniculatug were influenced much more by prior experience
with a habitat than were field stocks. This finding, if it applies to
fish, suggests that a field stock of '_I'_. mossambica might have been less

influenced by early habitat experience.

Applied Asmcts 9f lh_i_s_ Research

2. mossambica is an important culture fish in many parts of the
world (see Natural History Section). Research on habitat requirements
and mechanisms of habitat selection could be useful in the management of
this important species.

An understanding of fish behavior, and possible applications of
such knowledge, have largely been neglected in fisheries management
programs. One example is the put—and-take fishery (primarily trout).
Fish are reared in completely unnatural environments with no thought of
behavioral needs. hey are then placed in streams where they do not act
like wild fish. Hatchery-reared trout are very susceptable to hook-and-

line fishing, mortality is great, and over-wintering is poor. Perhaps,

_60

providing a meaningful environmental experience during a critical period
of their development would enable them to respond more like native fish.
If early habitat experience is as important to trout as it apparently is
to _T_. mossambica, perhaps only a short exposure to an adequate habitat
during the first few weeks would be sufficient to establish desirable
behavior of hatchery fish once they are released.

Two experiments (to my knowledge) have been conducted in an attempt
to ”improve" hatchery-reared trout to increase time at large, increase
spread of catch, and reduce mortality. Shetter and Cooper (1957)
reported and evaluated an experiment designed to train trout (brook,
rainbow, and brown trout) to (I) feed off the bottom, (II) feed off the
bottom and avoid predators, (III) feed off the bottom, avoid predators,
and use natural cover. The level III trained fish had better spread in
the anglers' creel over a period of days than did controls, but the
total return of fish was 15 to 20% less than the controls. Unfortunately,
most training was done on adult and not young fish. Training of
fingerling brook trout (level III) did not increase returns to anglers.
But a single planting of fingerling rainbows gave significantly greater
returns for level III trained fish over controls. Also, a single
planting of brown trout gave improved but not significant returns over
controls. It is unfortunate that the experiments with fingerlings were
not continued.

Thompson (1966) conditioned juvenile coho and chinook salmon to
avoid predators. Trained salmon had significantly reduced mortality.
Stomach analysis of predators collected from the study area showed two

and one-half as many untrained fish as trained fish in the stomach

 

61

contents. To my knowledge this study has not been published nor has the
procedure been ad0pted in hatchery Operations.

Due to the alteration of aquatic habitats by man, adequate habitat
for desirable fish is being reduced. At the same time the demand on
these aquatic habitats by sport fishing is increasing. The establishment
of sport species in marginal or submarginal habitats would be
advantageous. Perhaps by proper manipulation of heredity and/or early
experience, populations of desirable species could be established in

these habitats.

Hindsight fl 1332. m

The path behind us is often clearer than when it was ahead. In
other words, if I could start over again (with additional funds) I would
make some changes in the procedures. A more detailed look at each fish
in the experimental tanks would have been advantageous. I would suggest
the use of photocells and recording equipment to monitor the movement
between habitats. This would allow continuous recording, which would
reduce the variance between individuals (no 0% or 100% scores); allow
analysis of changes of habitat utilization during the course of a day
(e.g. before and after feeding); and reduce observer eye strain.
Behavioral measures would still have to be done by observation, but the
time devoted to each individual could be reduced. Also, I would.move
the behavioral observations into the first few hours of acclimation.
This would allow a better test of Sale's hypothesis. I believe these
few changes would have made the results ”cleaner".

The list of future experiments suggested.by this research is almost
endless. A few I consider most important fellow. One of the first

endeavors should be to determine if the results obtained here can be

“u. nf.’ m i- .— 1w

 

62

repeated, first with‘T, mossambica and then with other species. A field
investigation of habitat entrainment would be interesting, challpnging,
and crucial. The next step would be evaluation of habitat conditioning
' procedures for fisheries management programs.

Several other interesting lines of investigation are open. Of what
importance is habitat fixation in nest site selection of'T, mossambica?
How does social hierarchy effect habitat selection in'T.:mossambica? Of
what importance are cover, conspecifics, competition, predation, etc. in
habitat fixation and selection in.1.:mossambica? How can the effect of
habitat entrainment be strengthened (e.g. use of other pertinent habitat
cues)? How does stress (e.g. crowding or poor water quality) effect

habitat fixation and later selection?

 

1.

SUMMARY

Fish were conditioned (reared) to different habitats and later tested
for habitat selection, over a two-day period, in an experimental tank
offering a choice of the conditioned habitat (sand or stone) and an
alternate habitat (sand or stone). Three regimes of experience were
used (early, late, and continuous). Early-experience individuals
received experience with the conditioning habitat the first 60 days
of life, late-experience the last 60 days before testing, and
continuous-experience fish from the time of free-swimming to testing.
‘Within each experience regime, 20 fish conditioned to a habitat with
a sand substrate and 20 to a habitat with‘a stone substrate were
tested. Half were tested at four months of age and half at eight
months.

Early and continuous-experience fish (both four and eight months)
significantly selected the conditioned habitat on the second day
(but not the first) in terms of total time and number of habitat
visits. There was no significant difference between early and
continuous experience. There was an age difference for late-
experience fish on day two. Four-month fish showed no significant
effect of prior experience, but eight-month fish strongly avoided
the conditioned habitat on day two.

Early experience (before sexual maturity) was feund to be not only a
sufficient but also a necessary condition for habitat fixation in
Tilapia mossambica as judged by a short term selection of two days.
The relationship of stress and habitat selection is discussed in
light of the findings that: (1) selection was made on day two but

not on day one, (2) fish that fed on day two had a relatively strong

63

 

64

habitat selection, and (3) color patterns indicating fright were
relatively less common for fish visiting the conditioned habitat.
Several behavioral comparisons were made using swim time, number of
swim bouts, number of pivots, and number of substrate manipulations.
There were no definite differences in activity scores for the
comparisons of (1) day one vs. day two and (2) fish in conditioned
habitat vs. fish in unconditioned habitat. Fish visiting the sand
habitat did more swimming (time and bouts) and substrate manipulating
than fish in stone habitat. Stone-conditioned fish did more
swimming (time and bouts) and more substrate manipulating than sand-
conditioned fish. Likewise, fouremonth fish did more swimming (time
and bouts) and:more substrate manipulating than eight-month fish.
Pivoting scores were identical or similar for all comparisons, which
suggests that this behavior is unaffected by experience, age, or
habitat. Pivoting may be necessary for a fish to maintain its
"bearings” or awareness. The results of the behavioral comparisons
are discussed in relation to other studies.

The results are discussed in relation to the natural history of

T. mossambica and the possible application of this study to

fisheries management.

‘u-up L‘flk v“.‘_ 1-

 

 

LIST OF REFERENCES

Allanson, B. R., A. Bok, and N. I. van Wyk. 1971. The influence of
exposure toTlow temperature on Tilapia mossambica Peters (Cichlidae)
II. Changes in serum osmolarity, sodium and chloride ion
concentrations. J. Fish. Biol. 3:181-185.

Allanson, B. R. and R. G. Noble. 1964. The tolerance of Tilapia
mossambica (Peters) to high temperatures. Trans. Amer. Fish. Soc.

93:323-332-

Aronson, L. R. 1951. Orientation and jumping behavior in the gobiid
fish, Bathygobius soEorator. Amer. Mus. Novitates 1486:1-21.

Badenhuizen, T. R. 1967. Temperatures selected by Tilapia mossambica
(Peters) in a test tank with a horizontal temperature gradient.
Hydrobiologia 30(3-h):5#1-55#.

Baerends, G. P. and J. M. Baerends-Van Roon. 1950. An introduction to
the study of the ethology of cichlid fishes. Behaviour(suppl. 1):
1-2h2.

Bailey, R. M., J. E. Fitch, E. S. Herald, E. A. Lachner, C. C. Lindsey,
C. R. Robins, and W. B. Scott. 1970. A list of common and
scientific names of fishes from the United States and Canada (third
edition). Amer. Fish. Soc. Spec. Publ. No. 6. 150 p.

Baker, M. C. 1971. Habitat selection in four spine sticklebacks
(Apeltes guadracus). Amer. Midl. Natur. 85(1):239-242.

Baldes, R. J. and R. E. Vincent. 1969. Physical parameters of
microhabits occupied by brown trout in an experimental flume.
Trans. Amer. Fish. Soc, 98(2):230-238.

Beach, F. A. and J. Jaynes. 195h. Effects of early experience upon the
behavior of animals. Psychol. Bull. 51:239-263.

Behrend, E. R. and M; E. Bitterman. 196h. Avoidance-conditioning in
the fish: further studies of the CS-VS interval. .Amer. J. Psychol.

Berlyne, D. E. 1966. Curiosity and exploration. Science 153:25-33.

Boltt, R. E. Personal communication. Rhodes University, Institute for

Freshawater Studies, Dept. of Zoology and Entomology, Grahamstown,
South Africa. ‘

65

66

Bouck, G. R. and R. C. Ball. 1966. Influence of capture methods on
blood characteristics and mortality in the rainbow trout (Salmo
gairdneri). Trans. Amer. Fish. Soc. 95:170-176.

Braasch, M. E. and P. W. Smith. 1965. Relationships of the tapminnows
Fundulus notatus and Fundulus olivaceus in the Upper Mississippi
River valley, COpeia 1965(1): £53.

Breder, C. M., Jr. 1950. Factors influencing the establishment of
residence in shells by tropical shore fishes. Zoologica 35:153-158.

Breder, C. M., Jr. 1954. Further studies on factors influencing the
reactions of tropical shore fishes to shells. Zoologica 39:79-86.

Brock, V. E. l95h. .A note on the spawning of Tilapia mossambica in sea
water. Copeia 1954(1):72.

Butler, R. L. and V. M. Hawthorne. 1968. The reactions of dominant
trout to changes in overhead artificial cover. Trans. Amer. Fish.

Soc. 97(1):37.41.

 

Chacko, P. I. and B. Krishnamurthi. l95fi. Observations on Tila ia
mossambica Peters in Madras. Jour. Bombay Nat. Hist. Soc. 52(213):

3&9-353-

Chavin, Welter and J. E. YOung. 1970. Factors in the determination of
normal serum glucose levels of goldfish, Carassins auratus L. Comp.
Biochem. Physiol. 33(3):629-653.

Chen, Tung-Pai. 1953. The culture of tilapia in rice paddies in Taiwan.
Chinese American Joint Commission on Rural Reconstruction, Fisheries
Series No. 2. 30 p.

Chimits, Pierre. 1955. Tilapia and its culture a preliminary
bibliography. FAO Fisheries Bulletin 8(1):1-33.

Clemens, H. P. and T. Inslee. 1968. The production of unisexual broods
by Tilapia mossambica sex-reversed with methyl testosterone. Trans.

Amer. Fish. Soc. 97(1):l8-21.

Copley, Hugh. 1958. Common freshwater fishes of east Africa. H. F.
and G. Witherby Ltd., London. 172 p.

Crass, R. S. Personal communication. Natal Parks, Game and Fish
Preservation Board, P. O. Box 662, Pietermaritzburg, Natal, South
Africa.

Crass, R. S. l96h. Freshwater fishes of Natal. Shuter and Shooter,
Pietermaritzburg. 167 p.

Day, D. S. and W. G. Pearcy. 1968. Species associations of benthic
fishes on the continental shelf and slope off Oregon. J. Fish. Res.
Ed. Canada 25(12)32665-2675.

67

Devambez, L. C. 1964. Tilapia in South Pacific. S. Pacif, Bull. 14(0):
27-28, 52.

Donnelly, B. G. 1969. A preliminary survey of Tila ia nurseries on
Lake Kariba during 1967/68. Hydrobiologia 3EZ2§:195-206.

Everest, F. H. 1969. Habitat selection and spatial interaction of
juvenile chinook salmon and steelhead trout in two Idaho streams.
Ph.D. Thesis. Univ. Idaho, Moscow. 77 p.

Forbes, S. A. 1907. On the local distribution of certain Illinois
fishes: an essay in statistical ecology. Ill. Lab. Nat. Hist.,
Bull. 7(8):273-303.

Fowler, H. 1965. Curiosity and exploratory behavior. MacMillan, New
York. 216 p.

Fryer, Geoffrey. Personal communication. Freshwater Biological
Association, Windermere Laboratory, The Ferry House, Far Sawrey,
Ambleside, westmorland, England.

Haines, T. A. and R. L. Butler. 1969. Responses of'yearling smallmouth
bass (Micropterus dolomieui) to artificial shelter in a stream
aquarium. J. Fish. Res. Ed. Canada 26(1):21-3l.

Hallam, J. C. 1959. Habitat and associated fauna of four species of
fish in Ontario streams. J. Fish Res. Ed. Canada 16:1h7-173.

Harris, V. T. 1952. An experimental study of habitat selection by
prairie and forest races of'the deermouse, Perogyscu51maniculatus.
Contr. Lab. Vert. Biol. Univ. Michigan 56:1-53.

Hasler, A. D. 1956. Influence of environment reference points on
learned orientation in fish (Phoxinus). Z. vergleich. Physiol.
38:303-310.

Hasler, A. D. 1966. Underwater Guideposts. Univ. of Wisconsin Press,
Madison. 155 p.

Hasler, A. D. and W. J.'Wisbyu 1958. The return of displaced largemouth
bass and green sunfish to a "home" area. Ecology 39:289-293.

Hinde, Robert. 1966. Animal behavior: a synthesis of ethology'and
comparative psychology. MoGraw-Hill, New York. 53h p.

Hofstede, A. E. and F. Botke. 1950. Tilapia mossambica as a factor in
malaria control in the vicinity of Djakarta. Landbouw 22:h53.h68.

Hoover, F. G. and J. A. St. Amant. 1970. Establishment of“ ilapia
mossambica Peters in Bard Valley, Imperial County, California.
Calif. Fish and Game 56(1):?0-71.

Hunsaker, Don, II and R. W. Crawford. 196h. Preferential spawning
behavior of the largemouth bass, Micropterus salmoides. Capeia
1964(1):240-2hl.

68

Jenkins, T. M., Jr. 1969. Social structure, position choice and
micro-distribution of two trout species (Salmo trutta and Salmo
airdneri) resident in mountain streams. Anim. Behav. Monogr. 2(2):
57-123.

Jubb, R. A. Personal communication. Albany Museum, Grahamstown, Cape,
South Africa.

Jubb, R. A. 1961. An illustrated guide to the freshwater fishes of the
‘ Zambezi River, Lake Kariba, Pungwe, Sabi, Lundi and Limpopo River.
Stuart Manning, Bulawayo. 167 p.

Jubb, R. A. 1967. Freshwater fishes of Southern Africa. A. A. Balkema,
Cape Town. 21+8 p.

Kamal Pasha, S. M. 1961+. The anatomy and histology of the alimentary
canal of a herbivorous fish Tila ia mossambica (Peters). Proc.
Indian Acad. Sci., B. 59340459.

King, J. A. 1958. Parameters relevant to determining the effect of
early experience-upon the adult behavior of animals. Psychol.
Bull. 55:%-58.

Klopfer, P. H. 1963. Behavioral aspects of habitat selection: the role
of early experience. Wilson Bull. 75(1):15-22.

Klopfer, P. H. 1965. Behavioral aspects of habitat selection: a
preliminary report on stereotypy in foliage preferences of birds.
Wilson Bull. 77(4):376-381.

Klopfer, P. H. 1967. Behavioral sterotypy in birds. Wilson Bull. 793
290.300.

Klopfer, P. H. 1969. Habitats and territories. Basic Books, Inc.,
New York. 117 p.

Klopfer, P. H. and J. P. Hailman. 1965. Habitat selection in birds,
p. 279.303. 3; D. s. Lehrman, R. A. Hinde, and E. Shaw [5d,]
Advances in the study of behavior vol. 1. Academic Press, New York.

Kwain, W. and H. R. McCrimmon. 1967. The behavior and bottom color
selection of the rainbow trout, Salmo gairdneri Richardson, exposed
to different light intensities. Anim. Behav. 15(1):75-78.

Lachner, E. A., C. R. Robins, and W. R. Courtenay. 1970. Exotic fishes
and other aquatic organisms introduced into North America.
Smithsonian Contributions to Zoology No. 59. 29 p.

le Roux, P. J. 1956. Feeding habits of the young of four species of
W. South African J. Sci. 53:33-37.

Luchins, A. S. and R. H. Forgus. 1955. The effect of differential post-
weaning environments on the rigidity of an animals behavior. J.
Genet. Psychol. 86:51-58.

69

MacArthur, R. H. and R. Levine. 196#. Competition, habitat selection
and character displacement in a patchy environment. Proc. Nat.
Acad. Sci. 51:1207-1210.

McCrimmon, H. and W. Kwain. 1966. Use of overhead cover by rainbow
trout exposed to a series of light intensities. J. Fish. Res. B.
Canada 23(7):983-99o.

anro, J. L. 1967. The food of a community of East African freshwater
fishes. J. Zool. Soc. London 151:389-415.

Munther, G. L. 1970. Movement and distribution of smallmouth bass in
the Middle Snake River. Trans. Amer. Fish. Soc. 99(1):h#-53.

Nagase, G. 196h. Contribution to the physiology of digestion in
Tilapia mossambica (Peters): digestiverenzymes and the effects of
diets on their activity. Z. vergl. Physiol. h9(3):270-28#.

Neil, E. H. 196#. An analysis of color changes and social behavior of
Tilapia mossambica. Univ. of Calif. Publ. in Zoology 75(1):1-58.

Neil, E. H. 1966. Observations on the behavior of Tila ia mossambica
(Pisces, Cichlidae) in Hawaiian ponds. Copeia 1966(1):50-5 .

Newton, G. and S. Levine. 1968. Early experience and behavior, the
psychobiology of development. Thomas, Springfield, Illinois. 785 p.

Norris, K. S. 1963. The function of temperature in ecology of the
percoid fish, Girella nigricans (Ayres). Ecol. Monogr. 33(1):23-62.

Otis, L. S. and J. A. Cerf. 1963. Conditioned avoidance learning in
two fish species. Psychol. Rep. 12:679-682.

Panikkar, N. K. and P. R. S. Tampi. 1954. On the mouth-breeding
cichlid, Tilapia mossambica Peters. Indian J. Fish. l(1/2):217-230.

Pienaar, U. deV. 1968. The freshwater fishes of the Kruger National
Park. National Parks Bd. of Trustees of the Rep. of South Africa.
82 p.

Pongsuwana, U. 1956. Production of Tilapia mossambica in an experimental
pond at Bangkhen Thailand. Proc. Indo-Pacific Fish. Council 6(2):
197-201.

Potts, W. T. W., M. A. Foster, P. P. Rudy, and G. P. Howells. 1967.
Sodium and water balance in the cichlid teleost, Tilapia mossambica.
J. Exp. Biol. 47:h61-470.

Price, Edward. 1970. Differential reactivity of wild and semi-domestic
deermice (Peromyecus maniculatus). Anim. Behav. l8(#):7h7-752.

Russell, B. C. 1969. A checklist of the fishes of Goat Island, North
Auckland, New Zealand, with an analysis of habitats and associations.
J. Auckland Univ. Field Club 15:105-113.

70

Ruwet, J. C. 1968. Familial behaviour of Tila ia (Pisces, Cichlidae)
and its implications. Nature 217(5132):977.

Sale, P. F. 1968. Influence of cover availability on depth preference
of the juvenile manini, Acanthurus triostegus sandvicensis. COpeia

l968(4):802-807.

Sale, P. F. 1969a. A suggested mechanism for habitat selection by the
juvenile manini Acanthurus triostegus sandvicensis Streets. Behaviour

35(1/2):27-44.

Sale, P. F. 1969b. Pertinent stimuli fer habitat selection by the

invenile manini, Acanthurus triostegus sandvicensis. Ecology 50

4):6l6-623.

Sale, P. F. 1971. Apparent effect of prior experience on a habitat
preference exhibited by the reef fish, Dasc llus aruanus (Pisces:
Pomacentridae). Anim. Behav. 19(2):251-256.

Sargent, T. D. 1965. The role of experience in the nest building of
the Zebra Finch. Auk 82(1):48-61.

Schuster, W. H. 1955. Fish culture in conjunction with rice cultivation.
WOrld Crops 7:11-14 and 67-70.

Seitz, A. 1948. Vergleichende Verhaltensforschung an Buntbarschen.
Zeit. Tierpsyohol. 6:202-325.

Sheppard, D. H., P. H. Kopfer, and H. Oelke. 1968. Habitat selection:
differences in stereotypy between insular and continental birds.
Wilson Bull. 80:452-457.

Shetter, D. S. and G. P. Cooper. 1957. Creel returns and survival of
hatchery trout trained by psychological research services. Michigan
Dept. of Obnservation, Institute for Fisheries Research Report
No. 1510. ’

Shomura, R. S. 1964. Effectiveness of tilapia as live bait for Skipjack
tuna fishing. Trans. Amer. Fish. Soc. 93(3):291-294.

Sills, J. B. 1970. A review of herbivorous fish for weed control.
Prog. Fish. Cult. 32(3):l58-l6l.

Sokal, R. R. and F. J. Rohlf. 1969. Biometry. The principles and
practice of statistics in biological research. W. H. Freeman and
Co., San Francisco. 776 p.

Solomon, K. and B. R. Allanson. 1968. The effects of exposure to low
temperatures on the serum protein of Tilapia mossambica Peters.
Comp. Biochem. Physiol. 25:485-492.

St. Amant, J. A. 1966. Addition of Tilapia mossambica Peters to the
California fauna. Calif. Fish and Game 52(1):54:55.

Swingle, H. S. 1960. Comparative evaluation of two tilapias as pond
fishes in Alabama. Trans. Amer. Fish. Soc. 89(2):l42-l48.

71

Thompson, R. B. 1966. Effects of predator avoidance conditioning on
the post-release survival rate of artificially propagated salmon.
Ph.D. Thesis. U. of Washington. 169 p.

Thys van den Audenaerde, Dirk F. E. 1968. An annotated bibliography of
Tilapia (Pisces, Cichlidae). NMsee Royal De L'Afrique Centrale-
Tervuren, Belgique. 406 p.

Tinbergen, L. 1960. The natural control of insects in pine woods. I.
factors influencing the intensity of predation by song birds. Arch.
Neerl. Zool. 13:265-343.

Tinley, K. L. 1964. Fishing methods of the Thonga tribe in N. E.
Zululand, and Southern Mbsambique. Lammergeyer (Pietermaritzburg)

339-39-

Trewavas, Ethelwyun. 1966. A preliminary review of fishes of the genus
Tilapia in the eastward-flowing rivers of.Africa, with proposals of
two new specific names. Rev. 2001. Bot. Afr. 74:394-424.

Tyler, J. C. 1971. Habitat preferences of the fishes that dwell in
shrub corals on the Great Barrier Reef. Proc. Acad. Natur. Sci.
123(1):1-26.

von Uexkull, Jacob. 1934. A stroll thr h the world of animals and
men, p. 5-80. .lE C. H. Schiller [;d. Instinctive behavior (1957).
International Univ. Press, New York.

‘Wecker, S. C. 1963. The role of early experience in habitat selection
by the prairie deermouse Peromyscus maniculatus bairdi. Ecol.
monogr. 33(4):307-325.

'Wecker, s. c. 1964. Habitat selection. Sci. Amer. 211(4):109-116.

‘Welcomme, R. L. 1964. The habitats and habitat preferences of the
young of the Lake Victoria Tilapia (Pisces, Cichlidae). Rev. 2001.
Bot. Afr. 70(1/2):l-28.

'Welker, W. I. 1961. An analysis of exploratory and play behavior in
animals, p. 175-226. ;p_D. w. Fiske and s. R. Maddi [Edg7 Functions
of varied experience. Dorsey Press, Homewood, Illinois.

werboff, J. and T. Lloyd. 1963. Avoidance conditioning in the guppy
(Lebistes reticulatus). Psychol. Rep. 12:615-618.

‘Wiens, J. A. 1970. Effects of early'experience on substrate pattern
selection in Rana aurora tadpoles. COpeia l970(3):543-548.

   

”‘ilfliitg‘ullflmmMeitfitiilfiiuimfiri‘lfi‘s