Genetics in aquaculture proceedings of the Sixteenth U.S.-Japan Meeting on Aquaculture, Charleston, South Carolina, October 20 and 21, 1987

[From the U.S. Government Printing Office, www.gpo.gov]

L ILI @TE

NOAA Technical Repoi-t NMFS 92 November 1990

Genetics in Aquaculture

Proceedings of the Sixkenth
U.S. -Japan Meeting on Aquaculture
Charleston, South Carolina
October 20 and 21., 1987

Ralph S. Svrjcek (editor)

U.S. Department of Commerce

SH11
.A44672
no.92

NOAA Technical Report NMFS

The major responsibilities of the National Marine Fisheries Service (NMFS) are to monitor and assess the abundance and geographic distribution
of fishery resources, to understand and predict fluctuations in the quantity and distribution of these resources, and to establish levels for their optimum
use. NMFS is also charged with the development and implementation of policies for managing national fishing grounds, development and enforcement
of domestic fisheries regulations, surveillance of foreign fishing off United States coastal waters, and the development and enforcement of international
fishery agreements and policies. NMFS also assists the fishing industry through marketing service and economic analysis programs, and mortgage in-
surance and vessel construction subsidies. It collects, analyzes, and publishes statistics on various phases of the industry.
The NOAA Technical Report NMFS series was established in 1983 to replace two subcategories of the Technical Reports series: "Special Scientific
Report-Fisheries" and "Circular." The series contains the following types of reports: Scientific investigations that document long-term continuing
programs of NMFS; intensive scientific reports on studies of restricted scope; papers on applied fishery problems; technical reports of general interest
intended to aid conservation and management; reports that review in considerable detail and at a high technical level certain broad areas of research;
and technical papers originating in economics studies and from management investigations. Since this is a formal series, all submitted papers receive
peer review and those accepted receive professional editing before publication.
Copies of NOAA Technical Reports NMFS are available free in limited numbers to governmental agencies, both Federal and State. They are also
available in exchange for other scientific and technical publications in the marine sciences. Individual copies may be obtained from: U.S. Department
of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161. Although the contents have not been copyrighted
and may be reprinted entirely, reference to source is appreciated.

69. Environmental quality and aquaculture systems: Proceedings of the Beverly M. Vinter. October 1989, 651 p.
thirteenth U.S.-Japan meeting on aquaculture, Mie, Japan, October 24-25,
1984, edited by Carl J. Sindermann. October 1988, 50 p. 81. Catch-per-unit-effort and biological parameters from the Massachu-
setts coastal lobster (Horwrus ameriranus) resource: Description and Trends,
70. New and innovative advances in biology/engineering with potential by Bruce T. Estrella and Daniel J. McKiernan. September 1989, 21 p.
for use in aquaculture: Proceedings of the fourteenth U.S.-Japan meeting
on aquaculture, Woods Hole, Massachusetts, October 16-17, 1985, edited 82. Synopsis of biological data on the cobia R"hycentron canadum (Pisces:
by Albert K. Sparks. November 1988, 69 p. Rachycentridae), by Rosalie Vaught Shaffer and Eugene L. Nakamura.
December 1989, 21 p.
71. Greenland turbot Reinhardliushippoglossoidesof the eastern Bering Sea
and Aleutian Islands region, by Miles S. Alton, Richard G. Bakkala, Gary 83. Celaphopods from the stomachs of sperm whales taken off Califor-
E. Walters, and Peter T. Munro. December 1988, 31 p. nia, by Clifford H. Fiscus, Dale W. Rice, and Allen A. Wolman. Decem-
ber 1989, 12 p.
72. Age determination methods for northwest Atlantic species, edited
by Judy Perittila and Louise M. Dery. December 1988, 135 p. 84. Results of abundance surveys of juvenile Atlantic and Gulf menha-
den, Brevoortia Vannus and B. patrunus, by Dean W. Ahrenholz, James F.
73. Marine flora and fauna of the Eastern United States. Mollusca: Guthrie, and Charles W. Krouse. December 1989, 14 p.
Cephalopoda, by Michael Vecchione, Clyde F. E. Roper, and Michael
J. Sweeney. February 1989, 23 p. 85. Marine farming and enhancement: Proceedings of the Fifteenth
U.S.-Japan Meeting on Aquaculture, Kyoto, Japan, October 22-23, 1986,
74. Proximate composition and fatty acid and cholesterol content of 22 edited by Albert K. Sparks. March 1990, 127 p.
species of northwest Atlantic finfish, by Judith Krzynowek, jenny Mur-
phy, Richard S. Maney, and Laurie J. Panunzio. May 1989, 35 p. 86. Benthic macroiatina and habitat monitoring on the continental shelf
of the northeastern United States. 1. Biomass, by Frank Steimle. Febru-
75. Codend selection of winter flounder Pseudopleuronectes amen@anus, by ary 1990, 28 p.
David G. Simpson. March 1989, 10 p.
87. Life history aspects of 19 rockfish species (Scorpaenidae: Sebastes) from
76. Analysis of fish diversion efficiency and survivorship in the fish return the Southern California Bight, by Milton S. Love, Pamela Morris, Mer-
system at San Onofte Nuclear Generating Station, by Milton S. Love, ritt McCrae, and Robson Collins. February 1990, 38 p.
Meenu Sandbu, Jeffrey Stein, Kevin T. Herbinson, Robert H. Moore,
Michael Mullin, and John S. Stephens Jr. April 1989, 16 p. 88. Early-life-history profiles, seasonal abundance, and distribution of
four species of clupeid larvae from the northern Gulf of Mexico, 1982 and
77. Illustrated key to the genera of free-living marine nematodes of the 1983, by Richard F. Shaw and David L. Drullinger. April 1990, 60 p.
order Enoplida, by Edwin J. Keppner and Armen C. Tarjan. July 1989,
26 p. 89. Early-life-history profiles, seasonal abundance, and distribution of
four species of carangid larvae off Louisiana, 1982 and 1983, by Richard
78. Survey of fishes and water properties of south San Francisco Bay, F Shaw and David L. Drullinger. April 1990, 37 p.
California, 1973-82, by Donald E. Pearson. August 1989, 21 p.
90. Elasmobranchs as living resources: Advances in the biology, ecology,
79. Species composition, distribution, and relative abundance of fishes systematics, and the status of the fisheries, edited by Harold L. Pratt Jr.,
in the coastal habitat off the southeastern United States, by Charles A. Samuel H. Gruber, and Toru Taniuchi. July 1990, 518 p.
Wenner and George R. Sedberry. July 1989, 49 p.
91. Marine flora and fauna of the northeastern United States, Echinoder-
80. Laboratory guide to early life history stages of northeast Pacific fishes, mata: Crinoidea, by Charles G. Messing and John H. Dearborn. August
by Ann C. Matarese, Arthur W. Kendall Jr., Deborah M. Blood, and 1990, 30 p.

NOAA Technical Report NMFS 92

Genetics in Aquaculture

Proceedings of the Sixteenth
U.S. -Japan Meeting on Aquaculture
Charleston, South Carolina
October 20 and 21, 1987

Ralph S. Svrjcek (editor),
Publications Unit
Northwest and Alaska Fisheries Science Centers

Panel Chairmen:
Conrad Mahnken, United States
Takeshi Nose, Japan

Under the U.S. -Japan Cooperative Program
in Natural Resources (UJNR)

November 1990

stil OF Co. U.S. DEPARTMENT OF COMMERCE
Robert Mosbacher, Secretary
National Oceanic and Atmospheric Administration
John A. Knauss, Under Secretary for Oceans and Atmosphere
14, National Marine Fisheries Service
"ArEs 0V William W. Fox Jr., Assistant Administrator for Fisheries

C"
TaE7

LIBRARY
NOAA/CCEH
1990 HOBSON AVE.
CHAS. SC 29408-2623

PREFACE

The United States and Japanese counterpart panels on aquaculture were formed in 1969 under
the United States-Japan Cooperative Program in Natural Resources (UJNR). The panels
currently include specialists drawn from the federal departments most concerned with
aquaculture. Charged with exploring and developing bilateral cooperation, the panels have
focused their efforts on exchanging information related to aquaculture which could be of benefit
to both countries.
The UJNR was begun during the Third Cabinet-Level Meeting of the joint United
States-Japan Committee on Trade and Economic Affairs in January 1964. In addition to aqua-
culture, current subjects in the program include desalination of seawater, toxic microorganisms,
air pollution, energy, forage crops, national park management, mycoplasmosis, wind and
seismic effects, protein resources, forestry, and several joint panels and committees in marine
resources research, development, and utilization.
Accomplishments include: Increased communication and cooperation among technical
specialists; exchanges of information, data, and research findings; annual meetings of the panels,
a policy-coordinative body; administrative staff meetings; exchanges of equipment, materials,
and samples; several major technical conferences; and beneficial effects on international
relations.

Conrad Mahnken - United States
Takeshi Nose Japan

The National Marine Fisheries Service (NMFS) does not approve, recom-
mend or endorse any proprietary product or proprietary material mentioned
in this publication. No reference shall be made to NMFS, or to this publica-
tion furnished by NMFS, in any advertising or sales promotion which would
indicate or imply that NMFS approves, recommends or endorses any pro-
prietary product or proprietary material mentioned herein, or which has
as its purpose an intent to cause directly or indirectly the advertised pro-
duct to be used or purchased because of this NMFS publication.

Text printed on recycled paper

CONTENTS

W.K. HERSHBERGER Assessment of inbreeding and its implications for salmon broodstock 1
J.M. MYERS development
R.N. IWAMOTO
W.C. McAULEY

G.H. THORGAARD Chromosome set manipulation in salmonid fishes 9

R.T. DILLON Jr. Outcrossed lines of the hard clam Mercenaria mercenaria 11
J.J. MANZI

Y. FU A preliminary study on genetics of two types of the rotifer Brachionus plicatilis 13
Y. NATSUKARI
K. HIRAYAMA

K.FUKUSHO Present status of genetic studies on marine finfish in Japan 21

J.C. LEONG Recombinant viral vaccines in aquaculture 27
R. BARRIE
H.M. ENGELKING
J. FEYEREISEN-KOENER
R. GILMORE
J.HARRY
G.KURATH
D.S. MANNING
C.L. MASON
L.OBERG
J. WIRKKULA

R.S. WAPLES Genetic monitoring of Pacific salmon hatcheries 33
G.A. WINANS
F.M. UTTER
C. MAHNKEN

Sj. YOON Successful gene transfer in fish 39
Z. LIU
A.R. KAPUSCINSKI
P.B. HACKETT
A.FARAS
K.S. GUISE

T. NAKANISHI Clonal ginbuna crucian carp as a model for the study of fish immunology 45
H.ONOZATO and genetics

T.I.J. SMITH Aquaculture of striped bass, Morone saxatilis, and its hybrids in North America 53

L.J. LESTER Computerized image analysis for selective breeding of shrimp: a progress 63
K.S. LAWSON report
Mj. PIOTROWSKI
T.-C. B. WONG

H. MOMMA Breeding test on abalone 71

L.L. BEHRENDS Two-stage hybridization and introgression for improving production traits of 77
J.G. KINGSLEY red tilapias
A.H. PRICE III

Assessment of Inbreeding and Its Implications for
Salmon Broodstock Development*

WILLIAM K. HERSHBERGER and JAMES M. MYERS

School of Fisheries WH-10
University of Washington
Seattle, WA 98195

R.N. IWAMOTO** and W.C. McAULEY

Domsea Farnis, Inc.
5500 180th S. W.
Rochester, WA 98579

ABSTRACT

Inbreeding is an important part of any selection and breeding program designed to improve
aquacultural broodstock. A decrease in freshwater and saltwater growth rate was noted in a strain
of coho salmon, Oncorhynchus kisutch, undergoing selection to improve these traits for commercial
production. Thus, an investigation was undertaken to estimate the level of inbreeding in this
strain and to assess different approaches to alleviate problematic levels of inbreeding. Estimation
of inbreeding level was conducted via pedigree analysis and change in heterozygosity of
elctrophoretically detected ser-um proteins variants of odd- and even-year lines of coho salmon.
The two methods of analysis indicated vastly different inbreeding levels. However, pedigree
analysis, the more accurate of the two methods, estimated inbreeding levels not anticipated to
cause the observed depression in growth traits. Two approaches, interstock crosses and crosses
between parallel -selected lines, were assessed for alleviation of inbreeding problems. Both types
of crosses decrease the level of inbreeding, but the performance of the two types of crosses dif-
fered greatly. Crosses between unrelated year classes of the selected stock showed positive heterotic
effects, while the outcrosses with unrelated lines yielded negative heterotic effects. These results
indicate that careful attention should be given to the selection of the founding populations from
which broodstocks are developed and that subsequent breeding information be collected to pro-
duce pedigrees for population maintenance. Furthermore, the production of parallel "in-house"
lines, may provide the best method of minimizing inbreeding without diluting selection gains.

Introduction result in increased inbreeding levels (Falconer 1981), where
the magnitude will depend on the genetic characteristics
Inbreeding is integral to any selection and breeding pro- of the population and the severity of the constraints im-
gram designed for the development of broodstock. Such posed. Consequently, the factors that influence inbreeding
programs generally deal with a "closed" population (i.e., must be integrated into the design of any program to
migration into the population is eliminated) having a re- develop genetically improved aquacultural stocks.
stricted breeding population size. Both of these factors There has been a large amount of research concerning
inbreeding and its effects on various traits in fish. For
example, work with rainbow trout, Oncorhynchus mykiss
* Contribution No. 760, School of Fisheries WH-10, University of Wash- (formerly SaIrno gairdnert), has revealed that increased levels
ington, Seattle, WA 98195. The Project was supported by U.S. NOAA of inbreeding result in increased egg and fry mortality,
Grant NA86AA-D-SCO44 A09 to the Washington Sea Grant Program increased numbers of abnormal fry, decreased early
Project No. R/A-47.
* Current Address: Ocean Farms of Hawaii, P.O. Box A, Kailua-Kona, growth, and decreased fishery recovery (Kincaid 1976,
HI @6745 1983; Aulstad and Kittlesen 1971). Research with brook

2 NOAA Technical Report NMFS 92

trout, Salvelinusfontinalis, has demonstrated a negative im- a broodstock with traits that are beneficial to the produc-
pact on weight owing to inbreeding (Cooper 1961). Ryman tion of 300-350 g coho salmon for the "plate-size" salmon
(1970) reported a decrease in recapture frequency in Atlan- market.
tic salmon, Salmo salar, with increased levels of inbreeding. The traits that have been emphasized for selective im-
In general, the results of these studies suggest a negative provement are 1) freshwater growth, 2) smoltification, and
impact on a variety of biological traits in the populations 3) saltwater growth to harvest size. Genetic analyses of
studied and, consequently, on production. these traits in the stock employed by Domsea Farms re-
No studies have been published on the effects of inbreed- vealed adequate variability to expect progress from selec-
ing on Pacific salmon, Oncorhynchus spp., nor have any tion (Iwamoto et al. 1982; Hershberger and Iwamoto 1984;
published reports dealt with the effects of inbreeding in con- Saxton et al. 1984).
junction with a selection and breeding program designed Using estimated genetic values and considering that
to develop a genetically improved stock for aquacultural the facilities available to the program would only allow
purposes. To some degree, both of these deficiencies in in- raising 40 families of 600 individuals or less, a selection
formation are being eliminated as Pacific salmon are used scheme was designed to yield maximum response and to
for captive culture. It is imperative that data be obtained be useful in a commercial operation (Fig. 1). This scheme
on inbreeding in these species under defined programs to involved several different types of concurrent selection
determine their response to selection. (e.g., family and individual) and used a selection index that
incorporated heritability estimates, relative economic
values, genetic correlations, and mean values on all the
Research Rationale traits of interest. It was recognized early in the develop-
ment of this scheme that potential inbreeding problems
The University of Washington, Domsea Farms, Inc., and could arise from the rather severe limitation in breeding
the Washington Sea Grant Program have been conduct- population size (only twenty individuals contribute to
ing a selection and breeding program with coho salmon, each generation). Consequently, breeding was conducted
0. k1sutch, to develop a broodstock for the marine net-pen by a rotational line-crossing procedure (Fig. 2) to minimize
industry in the State of Washington. The major objective the possibility of crossing within lines. On a theoretical
of this nine year cooperative program has been to develop basis, these steps should limit the change in inbreeding

3.5 MONTH SALTWATER
SAMPLING

FRESHWATER
SAMPLING 8
MONTH
SALTWATER
SALTWATER SALTWATER SAMPLING
PHASEI PHASEII

Fish in Excess 25 Families*
of 600 per
Family*
FRESHWATER 15 Families
REARING SALTWATER
PHASE III

40 Families
20 Families*
INCUBATION MATURATION IN
FRESHWATER 14 MONTH
SALTWATER
SA
PH
'WATER
ASE

MATU RATI 0N IN
FRIES HWATER
S,
INCUBATION 'SAMPLING
10 Families
IN
PERFORMANCE Figure 1
60 Families 5 Families* Diagram of the selection scheme used to
develop coho salmon stocks for marine pen-
SPAWNING culture. The entire cycle represents a two-
year generation interval.

Hershberger et al.: Assessment of Salmon Broodstock Development 3

FAMILY FAMILY FAMILY FAMILY FAMILY
1. 2 3 4 5

FROM
FAMILY
5
TO
FAMILY1

FAMILY FAMILY FAMIL7 FAMIL7
6 7 8 9 10

FROM
FAMILY
10
TO Figure 2
FAMILY 6 Diagram of the rotational line mating
system used in crossing selected indi-
viduals. The asterisk indicates that
each family cross is composed of six
Q [9d lid E single-pair matings to form six double
I first-cousin families.

to about I % per generation (Hershberger and Iwamoto for this growth depression would be the accumulation of
1984). deleterious alleles through inbreeding. Even with the pre-
In 1983 (for the odd-year line) and 1984 (for the even- cautions taken in the design of the selection and breeding
year line) a decrease in the growth of selected fish in program, there were two potential sources of inbreeding
saltwater was observed (Fig. 3). One possible explanation that could not be quantitated. First, an unknown amount
of inbreeding may have been introduced by selection and
breeding that had occurred prior to use of this designed
program. Second, because of some unexpected husbandry
700 problems with raising fish to maturity there was a strong
probability that a few families contributed disporpor-
600 tionately to the subsequent generations. Prior to the defi-
nition of pedigrees for the two lines, the importance of these
500
Cr 2 factors was undeterminable.
400
................ . 2 As a result of these indications, studies were initiated
300 to 1) determine the actual levels of inbreeding in the two
lines and 2) define the best approach to eliminate inbreed-
200 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 ing in the selected stocks.
BROODYEAR

-M - ODD-YEAR LINE -0 EVEN-YEAR LINE -C- WILD CONTROLS
Determination of Inbreeding Level
Figure 3 The level of inbreeding in each of the two selected lines
Average weight (grams) of selected broodstock and wild controls.
after 8 months rearing in marine net-pens. Weights for 1986 are (Le., odd- and even-year) was determined by two differ-
given as unadjusted (1) and adjusted (2) for density differences ent methods. First, pedigree analyses were employed to
that year. N = 1200-2200 for selected broodstock and N determine the coefficient of inbreeding (F) (Falconer 1981).
15-35 for wild controls. Computation of this value is accomplished by tracing the

4 NOAA Technical Report NMFS 92

DOMSEA COHO SALMON
SEAWATER BROODSTOCK
ODD-YEAR LINE PEDIGREE BROODYEAR

K777N
-(R % 1977
LDTC LUU LAUC @AUC AU LD C RAT RUT RDU
,6 %
1 14 20 25 2 3 35 53 30

@rH A 1U @L'D
LHUS RHUU 'R@ PUC RHPU,' LHPUC C U LDPU DPU RDUU 1979
63 3 19 20,.,A 27 41 31 23 25
A

LOPU LH RDUUC R A
LDDU LHUSC RHDUC'@ LDAUC DDUC 1981
44 10 0 23 61 25
34 2

U
-(L' RHAUC RDUUC
'DD'U"', rLHUS RDD RDUU LDDUC' ,RDRUj: LDUU 1983
54
,66 59,@ 16 50..,@ 18 9 3

--- (L U S@) _S C'% LH 'U'S1 1985
@'_LDDUC RD RDDU
LHU rLHUU
LHUD S rL Z U LHPU %
54 52 2 11 41 55 50
A
@c ---

Figure 4
Pedigree of matings between se-
lected families for the odd-year
1987 broodstock line (1978-1986).
Families enclosed by a striped
box are double first cousins.

pedigree back to common ancestors and determining the ing coefficient from these pedigrees (Table 1) indicates that
probability that a pair of alleles, are identical by descent. the current level of inbreeding is not too severe, although
Second, the change in genotype frequencies of electro- the estimate for the next generation (1987 broodyear) will
phoretically analyzed protein differences were determined approach 8- 10%. These levels of inbreeding would not be
and the difference in heterozygote frequencies equated to anticipated to cause the level of change found in the re-
an apparent inbreeding coefficient (Hartl 1980). Electro- sponse of growth to selection. It has been estimated that
phoretic analyses were conducted on serum samples from in domesticated animals selection can balance an increase
100-120 adult fish in each of four years (1977, 1978, 1985, in inbreeding of approximately 2 % per generation (Pirch-
and 1986). The electrophoretic procedures employed were ner 1969). The estimated levels of inbreeding in coho
those reported in Utter et al. (1970) for analysis of serum salmon lines, to the point where apparent inbreeding
transferrins in coho salmon. depression was noted (1983 and 1984), are below this value.
Construction of the pedigrees for the two lines of coho However, the coefficients reflect only the inbreeding since
salmon revealed more closely related families than was the program was initiated and do not provide a measure
originally anticipated (Fig. 4). Calculation of an inbreed- of prior inbreeding. Further, it is difficult to determine what

Hershberger et al.: Assessment of Salmon Broodstock Development 5

Table 1
Inbreeding estimates based on pedigree analysis for both odd- and even-year lines, and
based solely on effective population size (Ne). The estimates are calculated assuming the
initial inbreeding coefficient (F) is equal to 0.

Pedigree estimates AF = (1 /2 N + 4)'

Odd-year Control Even-year Control Odd Even

1977 0.00 - 1978 0.00 - 0.00 0.00
1979 0.00 2.50 1980 0.00 2.50 2.27 2.27
1981 0.32 4.71 1982 0.63 5.75 4.49 4.41
1983 2.34 8.68 1984 4.22 9.11 8.78 6.58
1985 4.86 11.00 1986 5.90 12.20 10.86 8.52
1987 8.34 13.79 12.88

'Theoretical A F excluding sib-matings.

the effects of an incremental change in inbreeding may be much larger value than was obtained from the pedigree
in a species that has been recently developed from naturally analyses (Table 1).
reproducing populations (Soule 1980). It is possible to rationalize the discrepancy in these values
The second type of inbreeding assessment employed elec- on two bases. First, there is evidence suggesting selective
trophoretic analysis of the transferrin locus, which has been differences among the various afleles of the transferrin locus
shown to have three variant alleles (Utter et al. 1970) and (Suzurnoto et al. 1977; Pratschner 1977). T he results of
is one of the few genetically variable protein loci found in Pratschner's research indicated that fish with the "A" and
coho salmon (Utter et al. 1980). Comparison of the geno- "C" alleles, were more resistant to challenges by Vibrio
type and gene frequency values in the original adult pop- bacteria than those with the "B" allele, and Suzurnoto et
ulation with those from the fourth generation of selected al (1977) found that the "A" allele imparted higher sur-
stock (Table 2) revealed changes that would be anticipated vival to BKD (bacterial kidney disease) challenge. If such
in an inbred population (Falconer 1981); that is, there was selective pressures were applied to the selected coho salmon
a decrease in the frequency of heterozygotes and, with one lines, analyses based on the genotype frequencies would
exception, there was little change in the gene frequencies. tend to overestimate the inbreeding coefficient. The data
Calculation of apparent inbreeding coefficients based on from the current study support the hypothesis that fish with
the frequency change in heterozygotes (Fig. 5) yields a the "A" and "C" alleles have a selective advantage, and

Table 2
Observed transferrin gene and genotype frequencies in the odd- and even-year lines of
coho salmon and their changes over four generations of selection (N = 100- 120).

Odd-year broodstock line
Genotype Gene frequency

Year AA AB AC BB BC cc fA fB fC

1977 0.00 0.08 0.33 0.00 0.13 0.48 0.20 0.10 0.70
1985 0.05 0.03 0.08 0.00 0.18 0.68 0.10 0.10 0.80
Change +0.05 -0.05 -0.25 +0.00 +0.05 +0.20 -0.10 0.00 +0.10

Even-year broodstock line
Genotype Gene frequency

Year AA AB AC BB BC cc fA fB fC

1978 0.10 0.05 0.45 0.05 0.25 0.10 0.35 0.20 0.45
1986 0.12 0.00 0.42 0.00 0.04 0. 42 0.33 0.02 0.65
Change +0.01 -0.05 -0.02 -0.05 -0.21 +0.32 -0.02 +0.18 +0.20

6 NOAA Technical Report NMFS 92

Figure 5
ODD-YEAR Estimates of the apprent inbreeding coefficients for the odd- and
even-year lines based on the changes in observed and expected
ACTUAL HETEROZYGOSITY VS. EXPECTED genotype frequencies of the coho transferrin alleles.

1977 .525 -.46 /.46 = +14.1 %

1985 .275 -.525 /.525 = - 47.6 %
CHANGE IN OBSERVED HETEROZYGOSITY Table 3
The relative growth and survival of interstrain (Domsea
x Univ. of WA) and intrastrain (Domsea odd- x even-
CHANGE 1977 to 1985:.275 - .525 /.525 47.6 % year) crosses after 8 months rearing in marine net-pens.
ESTIMATED AF 41.9% The weights and survivals have been standardized against
the Domsea x Domsea (2 x 2) cross = 100. The index
value is the cross-product of weight and survival/100. N
EVEN-YEAR 8-45 for each cross.
Outcrossing schemes
ACTUAL HETEROZYGOSITY VS. EXPECTED Relative Relative
weight survival Index
1978 .75 -.635 /.635 = + 18.1 %
DOMSEA (D) 100 100 too
1986 .461 - .465 /.465 0.8 % D x UW (9 x 0) 147 25 36.8
CHANGE IN OBSERVED HETEROZYGOSITY UW x D (9 x a) 141 25 35.3
University of Wash. (UW) 55.1 21.4 11.7
CHANGE 1978 to 1986:.461 -.75 /.75 38.5 % DOMSEA line crosses
ESTIMATED AF 29.3% Relative Relative
weight survival Index

DOMSEA (2 x 2) 100 100 too
DOMSEA (2 x 3) 116.1 150 174.5
vibriosis is a common problem in the marine net-pen DOMSEA (3 x 2) 101.4 225 174.1
culture of salmon. The directed selection practiced on the DOMSEA (3 x 3) 128.7 100 128.7
stock may also have an epistatic effect on the transferrin
locus. A tacit assumption made in the use of the genotype
frequency relationship used to calculate an inbreeding coef-
ficient is the absence of selection. Such an assumption is coho salmon stock and the hatchery stock of the Univer-
clearly not valid in this situation and may result in the sity of Washington, and between the Domsea odd- and
inflation of the calculated value. even-year parallel- selected lines. Progeny from these crosses
To summarize, it appears that pedigree analysis is the were reared in conjunction with the broodstock line.
best approach to determine inbreeding levels in coho It is apparent from the data (Table 3) that the progeny
salmon. Thus, it would seem wise to assure that a selec- from the crosses derived from the Doinsea intrastock crosses
tion and breeding program incorporates the mechanisms were superior to the interstock cross at the time of harvest.
that define accurate pedigrees of the breeding population. Although both of the University of Washington x Domsea
Further, caution should, be exercised in the use of geno- hybrids were larger after eight months of saltwater rear-
type frequency changes to determine absolute values for ing, relative to the Domsea controls, the overall survival
inbreeding coefficients. The potential effects of direct and of both the hybrids and the University of Washington fish
indirect selection must be determined for these values to was extremely poor under net-pen conditions. The high
be considered as valid measurements of inbreeding. values reported reflect the survival of a few large hybrids
which biased the weight measurements. The University of
Washington x Domsea hybrids may not necessarily be
Elimination of Inbreeding indicative of all interstock crosses, but the results suggest
that extensive hybrid testing may be necessary to identify
Although the apparent levels of inbreeding in the selected a complementary stock. The Domsea intrastock hybrids,
stocks of coho salmon were not large, two approaches to however, showed both good growth and greatly improved
elimination of accumulated inbreeding were investigated: survival relative to controls. Maintaining "in-house"
outcrossing between stocks and outcrossing between lines parallel selection lines may be a more efficient expenditure
within stocks. Test crosses were made between the Domsea of effort relative to testing outcrosses. The "odd x even"

Hershberger et al.: Assessment of Salmon Broodstock Development 7

crosses would appear to be the method-of-choice for allevi- Citations
ating the inbreeding "load" while preserving selection
gains. Aulstad, D., and Kittlesen.
1971. Abnormal body curvatures of rainbow trout (Salmogairdnen)
inbred fry. J. Fish. Res. Board Can. 28:1918-1920.
Cooper, E.L.
Implications for 1961. Growth of wild and hatchery strains of brook trout. Trans.
Am. Fish. Soc. 23:614-617.
Broodstock Development Falconer, D.S.
1981. Introduction to quantitative genetics. Longman,Inc.New
The coho salmon stocks that have been developed as a York, NY, 340 p.
result of this research program have, apparently, not yet Gall, G.A.E.
1987. Inbreeding. In Population genetics & fishery management
reached a level of inbreeding which would result in a strong (N. Ryman and F. Utter, eds.), p. 47-87. Univ. Washington
negative impact on their performance. The depression in Press, Seattle, WA.
growth observed in both lines appears to have been envi- Hart], D.L.
ronmentally generated and subsequent generations have 1980. Principles of population genetics. Sinauer Assoc., Inc.
Sunderland, MA, 488 p.
performed well (Fig. 3). However, analyses of inbreeding Hershberger, W.K., and R.N. Iwamoto.
in these lines have demonstrated several areas requiring 1984. Systematic genetic selection and breeding in salmonid culture
special consideration in the development of aquaculture and enhancement programs. In Proceedings of the I lth U.S.-
broodstocks. Where possible, a selection and breeding pro- Japan Meeting on Aquaculture, salmon enhancement; 19-20
gram should be initiated with a large enough population October 1982, Tokyo, Japan, p. 29-32. U.S. Dep. Commer.,
NOAA Tech. Rep. NMFS 27.
size to completely address the combined needs of a reason- Iwamoto, R.N., A.M. Saxton, and W.K. Hershberger.
able selection differential and elimination of close familial 1982. Genetic estimates for length and weight of coho salmon
relationships. Otherwise, definitive steps must be taken in (Oncorhynchus kisutch) during freshwater rearing. J. Hered.
the formulation of the selection and breeding program to 73:187-191.
minimize the accumulation of inbreeding from these Kincaid, H.L.
1976. Effects of inbreeding on rainbow trout populations. Trans.
factors. Am .Fish. Soc. 105:273-280.
Further, a broodstock program should be initiated from 1983. Inbreeding in fish populations used for aquaculture. Aqua-
either an outbred population with an inbreeding coeffici- culture 33:215-227.
ent (F) equal, or close to 0, or from a stock with a defined Pirchner, F.
and well maintained pedigree. This would insure that the 1969. Population Genetics in Animal Breeding. W.H.Freeman
and Co., San Francisco, CA, 274 p.
inbreeding level could be unquestionably determined and Pratschner, G.A.
the effects of any increases could be well defined. In addi- 1977. Relative resistance of six transferrin phenotypes of coho
tion, research is needed to determine the response of aqua- salmon to cytophagosis, furunculosis and vibriosis. M.S. Thesis,
cultural species recently derived from wild populations to Univ. Washington, Seattle, WA, 71 p.
an increase in inbreeding level. While the response of Ryman, N.
domesticated animals to increases in inbreeding has been 1970. A genetic analysis of recapture frequencies of released young
of salmon (Salmo salar L.). Hereditas 65:159-160.
quantitated to some degree (Pirchner 1969), there is no Saxton, A.M., W.K. Hershberger, and R.N. Iwamoto.
a priori method by which to predict the magnitude of 1984. Smoltification in the net-pen culture of accelerated coho
responses in natural populations. As indicated by Gall salmon (Oncorhynchus kisutch); quantitative genetic analysis. Trans.
(1987), the best information will be obtained from induc- Am. Fish. Sec. 113:339-347.
Soule, M.E.
ing high levels of inbreeding in such stocks and quantify- 1980. Thresholds for survival: maintaining fitness and evolutionary
ing the effects. However, inbreeding effects observed in potential. In Conservation Biology (M.E. Soule and B.A. Wilcox,
the progeny of sib-matings are indicative of, but not highly eds.), p. 151-169. Sinauer Assoc., Inc., Sunderland, MA.
correlated with the performance of individuals with equal Suzumoto, B. K., C. B. Schreck, and J. D. McIntyre.
inbreeding levels produced through generations of matings. 1977. Relative resistances of three transferrin genotypes of coho
salmon (Oncorhynchus kistuch) and their hernatological responses to
Finally, it appears that using parallel selection in at least bacterial kidney disease. J. Fish. Res. Board Can. 34:1-8.
two separate lines of broodstock would be a valuable ap- Utter, F.M., W.E. Ames, and H.O. Hodgins.
proach to incorporate into a selection and breeding pro- 1970. Transferrin polymorphism in coho salmon (Oncorhynchus
gram. This provides an additional data set with which to kisulch). J. Fish. Res. Board Can. 27:2371-2373.
evaluate a selection program and also incorporates a Utter, F. M., D. Campton, S. Grant, G. Milner, J. Seeb, and L. Wishard.
mechanism that has the potential to eliminate inbreeding 1980. Population structures of indigenous salmonis species of the
Pacific Northwest. In Salmonid ecosystems of the North Pacific
effects without the loss of advances made in the traits that (W.J. McNeil and D.C. Hirnsworth, eds.), p. 285-304. Oregon
are beneficial to aquaculture production. State Univ. Press, Corvallis, OR.

Chromosome Set Manipulation in Salmonid Fishes

GARY H. THORGAARD

Department of Zoology and Program in Genetics and Cell Biology
Washington State University
Pullman, WA 99164-4220

ABSTRACT

Techniques to manipulate chromosome sets and produce polyploid fishes or fishes with all the
inheritance from the female or male parent have been exploited in aquaculture in recent years.
Some of the principal applications of this work have been to produce sterile fish or to produce
monosex populations. Three additional applications of chromosome set manipulation that we
have explored in salmonids in our laboratory and in collaboration with other laboratories have
been 1) increased survival in triploid hybrids; 2) the potential for gene transfer by "incomplete
gynogenesis"; and 3) the generation of homozygous diploids and ultimately homozygous clones
through androgenesis (all-paternal inheritance).
A number of researchers have demonstrated that interspecific triploid fish hybrids survive better
than the corresponding diploid hybrids. Notable examples of this phenomenon include the tiger
trout (brown trout x brook trout) hybrid, the rainbow trout x coho salmon hybrid, and the
chum salmon x chinook salmon hybrid. The tiger trout has considerable potential as a sport
fish and may be advantageous because both the diploid and triploid hybrids are essentially sterile.
The rainbow trout x coho salmon hybrid has increased resistance to IHN (infectious hernatopoietic
necrosis) virus characteristic of the coho salmon parent. The chum salmon x chinook salmon
hybrid has early seawater tolerance characteristic of the chum salmon parent.
Gynogenesis (all-maternal inheritance) experiments have normally involved complete inactiva-
tion of the paternal genome by radiation or chemical treatment of the sperm. However, we have
demonstrated that if a lower than normal radiation treatment is applied to the sperm, some paternal
genes may still be active in the progeny. This has been demonstrated for both pigmentation and
isozyme loci. It appears that the paternal genes in this situation are located on chromosomal
fragments which are lost during development. If the paternal genes can be stably inherited and
if desirable paternal traits can be selected for, this "incomplete gyogenesis" might potentially
be used to transfer desirable traits between species.
Androgenesis is induced by fertilizing radiation-inactivated eggs with normal sperm and by
applying a pressure or heat treatment to block the first cleavage division and produce homo-
zygous diploids. We have successfully induced androgenesis in rainbow trout and have also pro-
duced androgenetic progeny from homozygous androgenetic males. Androgenesis has a number
of distinctive applications for aquaculture, including generation of homozygous clones and recovery
of strains from cryogenically preserved sperm.

Outcrossed Lines of the Hard Clam Mercenaria mercenaria

ROBERT T. DILLON Jr.

Department of Biology
College of Charleston
Charleston, SC 29424

JOHN J. MANZI

Marine Resources Research Institute
Charleston, SC 29412

ABSTRACT

A large-scale breeding program has been initiated in South Carolina to achieve improved growth
and survival of the hard clams, M. mercenaria. This interdisciplinary, multi-institutional program
uses the facilities and personnel of the South Carolina Wildlife and Marine Resources Research
Institute, the College of Charleston, the University of South Carolina, and Clemson University.
Nursery stocks of hard clams that had been selected for fast growth were obtained from
Aquaculture Research Corporation ("ARC" - Dennis, MA) and the Virginia Institute of Marine
Science ("VIMS" - Wachapreague, VA). These stocks were compared to corresponding wild
populations for allele frequencies at seven polymorphic enzyme loci. Although as few as 30-60
parents were spawned at each of four generations to produce these two broodstocks, neither line
exhibited any reduction in heterozygosity. Both lines, however, showed evidence of genetic drift
and loss of rare alleles, suggesting that crosses between them could result in genetically distinct lines.
ARC and VIMS stocks were spawned on three occasions at different times of the year for pro-
duction of both reciprocal outbred and pure control lines. Growth and survival were monitored
regularly over two years. Early growth was strongly influenced by time of spawning, and as such
was not a reliable indicator of subsequent growth. Most significant disparities between trials
decreased as the lines aged. At 21 months, outbred and purebred lines were not consistently
different in their heterozygosity, mean size, or size variance.
Within crosses, little relationship was detected between shell length and heterozygosity aver-
aged over the seven enzyme loci. However, significant differences between the largest and smallest
clams were detected at individual loci in 10 of 42 tests. Results were consistent neither with the
hypothesis that the alleles themselves were affecting growth, nor with the hypothesis that these
enzyme loci were tightly linked to other loci affecting growth. Rather, it appears that alleles are
marking the entire genomes of their parents, and that variation in the growth rates of the off-
spring from individual clams may be obscuring any relationship with overall heterozygosity.

A Preliminary Study on Genetics of Two Types of the Rotifer
Brachionus plicatilis

YONG FU, YUTAKA NATSUKARI, and KAZUTSUGU HIRAYAMA

Faculty of Fisheries
Nagasaki University
Bunkyomachi, Nagasaki
Nagasaki 852, Japan

ABSTRACT

The domesticated rotifer Brachionus plicatilis can be divided roughly into two types, called
L and S, using morphological differences in the shape of anterior spines on the lorica (obtuse
angled and pointed, respectively). However, differences in growth responses with respect to
environmental factors make this method unreliable, We have, therefore, tried to clarify differ-
ences at the genetic level between types, using starch gel electrophoresis of enzymes.
Thirty-four collected strains were separated by three methods into the two types. Initially, strains
were qualitatively judged with respect to differences in the shape of anterior spines. Afterwards
pure strains were cultured parthenogenetically and re-evaluated using the second method (quan-
titative). To accomplish this, morphological features were measured, the ratios of which created
an index for comparison of the strains (cluster analysis). Both the anterior spine and cluster analysis
indicated that the 34 strains were composed of two large clusters consisting of 15 L and 19 S strains.
Allozyme variations of the 34 strains were then detected by horizontal starch gel electrophoresis.
Nine isozyme loci were recognized. Of the 42 alleles observed, 15 alleles over 6 loci showed great
differences between L- and S-types. Using genetic distances according to the allele frequencies
of 42 alleles, a dendrogram was drawn. The strains separated into two groups. One group con-
sisted of only S-type strains, the other group was subdivided again into 3 clusters. One of these
three clusters consisted only of the S-type strains, while the other two contained only L-type strains.
This result indicates the great genetic differences between L and S strains.

Introduction not to genetic differences. However, Fukusho and Okauchi
(1982, 1983, 1984) have provided evidence that differences
Since the introduction of the rotifer Brachionus plicatilis to may be genetic and that the two types can be isolated from
nourish larval fish, aquaculturists have increased scientific each other. In countries outside Japan, many scientists
attention on this organism. In Japan a significant achieve- recognize the variation of rotifers which is due to poly-
ment in rotifer biology was the discovery that the domes- morphosis. Scientific approaches concerning analysis of
ticated rotifers can be divided roughly into two so-called allozyme variation have therefore been investigated (Serra
S and L types as shown in Figure 1 (Fukusho 1983). The and Miracle 1983, 1985, 1987; Snell and Carrillo 1984;
main morphological differences between the two types are Snell and Winkler 1984; Suzuki 1983, 1987; King and
lorica size, lorica shape, and the shape of the anterior spines Zhao 198 7), while in Japan there have been no studies to
on the lorica. They also exhibit differences in growth with detect allozyme variation in the two types by means of elec-
respect to temperature. The morphological and physio- trophorctic procedures.
logical differences in the two types were summarized in a Using strains collected from many locations, we at-
previous review (Hirayama 1987). The rotifer, especially tempted to distinguish L and S types using morphological
the domesticated rotifer, exhibits cyclomorphosis (seasonal comparisons. In order to confirm the genetic differences
variation in size) and also polymorphosis (change in size between strains, allozyme variations were detected by
influenced by variations in diet) (Fukusho and Iwamoto horizontal starch gel elect rophores is. Then, the genetic
1980, 1981). So, there is a probability that observed dif- distances among collected strains were compared for mor-
ferences could be attributed to cyclo- or poly-morphosis, phological similarities.

14 NOAA Technical Report NMFS 92
S
N. @A
*@w
T 7 Ai

IVV

Figure 1
The two types of rotifer Brachionus plicatilis, L and S (provided by K. Fukusho).

Materials and Methods collected those eggs into test tubes reculturing them again
with marine Chlorella. After the offspring hatching from
We collected many strains from all over the world. On the those eggs grew and laid their first eggs, we performed mor-
map (Fig. 2), the localities of 34 strains used in this study phological measurements. We removed 20 individuals per
are shown. Table I shows the abbreviated names and sample and measured seven morphological features (Fig.
origins of the strains. In the tables and figures, L- and 3, A through G). The ratios of these measurements were
S-type strains are shown by abbreviation with capital and used to create indices for a cluster analysis.
small letters, respectively.
Allozyme Analysis
Morphological Analysis The same 34 strains were used both for electrophoretic and
We first observed the anterior spines of each of the 34 morphological analysis. Allozyme analysis for each strain
strains and qualitatively divided them into the two types, was conducted with a population grown from one in-
L and S, according to whether they had obtuse angled or dividual and cultured with marine Chlorella and baker's
pointed spines, respectively. We classified 15 strains into yeast. The population was harvested with a net, washed
the L type and 19 strains into the S type. After the initial with clean seawater several times, blotted dry using filter
screening, one individual from each strain was selected for paper and frozen at 30'C until analyzed. Before har-
culturing parthenogenetically and was regarded as one vesting, the group was starved for one day to remove the
genetic strain for further study. Each strain was cultured influences of food. Immediately prior to electrophoretic
with marine Chlorella (Nannochloropsis oculata). We collected analysis, we thawed the sample and used a small amount
eggs and recultured each strain in marine Chlorella suspen- of the drip absorbed by filter paper as a crude extract of
sions in 23'C. The first eggs were laid after 48 hours. We enzyme for allozyme alayisis. Electrophoresis were carried

Fu et al.: Genetics of the Rotifer Brachionus plicatilis 15

F-PA
F-PA- I I
F-PA- I I I
F-PA-IV

a-sal

.0 c _xM a-mk

P-ilo
t-son p-le -amp
is-eil t-Pu -0t

s-sin

i-ja

i-kay
J-SAP
J-NSGT J-SAP-86
J-NSGT-11 J-TKU

J-NSU
J-NSC 0@4
J-NSZ
i-nsz j-ebp J_S01
i-kgko i-otk
j-kgko-8 J-OTK
J-KAU J-kgs locality of collection

Figure 2
Map of collection localities. Capital and small letters mean L- and S-type strains, respectively.

out in 11% starch gel with three buffer systems reported Leucine aminopeptidase (LAP, EC 3.4.11. 1); and Glucose
by Clayton and Tretiak (1972) with minor modifications phosphaste isomerase (GPI, EC 5.3.1.9).
(Table 2). Staining procedures were from Shaw and Prasad
(1970) and Siciliano and Shaw (1976). The following 18
enzymes were tested: a-Glycerophosphate clehydrogenase Results
(aGPD, EC 1. 1. 1. 8); D-Sorbitol dehydrogenase (SDH, EC
1.1.1.14); Lactate dehydrogenase (LDH, EC 1.1.1.27; Morphological Analysis
3-Hydrooxybutyrate dehydrogenase (HBDH, EC
1.1.1.30); Malate clehydrogenase (MDH, EC 1.1.1.37); In Figure 3, are shown the average morphological mea-
Malic enzyme (ME, EC 1.1.1.40); Isocitrate dehydro- surements and standard deviations of the 15 L and 19 S
genase (IDH, EC 1.1.1.42); 6-Phosphogluconate strains. The results indicate that the strains of the rotifer
dehydrogenase (6PGD, EC 1. 1. 1. 44); Glucose-6-phosphate could be divided clearly into the two types by quantitating
4

dehydrogenase (G6PD, EC 1.1.1.49); Superoxide the shape of the anterior spine (E/D, G/F). The results of
dismutase (SOD, EC 1. 15. 1. 1); Aspartate aminotrans- the cluster analysis (Fig. 4A) are identical to the classifica-
ferase (AAT, EC 2.6. 1. 1); Adenylate kinase (AK, EC tion judging by the anterior spine shape (15 L types, 19
2.7.4.3); Phosphoglucomutase (PGM, EC 2.7.5. 1); S types). Each cluster can be divided again into 2 small
Esterase (EST, EC 3. 1. 1. 1); Alkaline phosphatase (ALP, clusters. These results indicate that with statistical treat-
EC 3.1.3. 1); Acid phosphatase (ACP, EC 3.1.3.2); ment of the morphologica charcteristics, the varieties of the

16 NOAA Technical Report NMFS 92

Table I
Abbreviated names and origins of 34 strains of Brachionas plicatilis tested for morphological and genetic differences. PE: Prefec-
tural Experimental Station or Hatchery; SFC: Japan Sea Farming Center; AQD SEAFDEC: Aquaculture Division of South
East Asian Fisheries Development Center; NICA: National Institute of Coastal Aquaculture; and CE: City Hatchery. Capital
and small letters mean that the strain belongs to L and S type respectively.

Abbreviated Station or Year of Wild (w) or
name Country Locality hatchery collection domesticated (d)

j-amp Japan Aomori PE '87 d
j-kay Japan Kagawa PE '87 d
j-ehp Japan Ehime PE '87 d
j-otk Japan Oita SFC '87 d
j-nsz Japan Nagasaki PE '86 d
j-kgko Japan Kagoshima (Kai Lake) '78 w
j-kgko-86 Japan Kagoshima (Kai Lake) '86 w
j-kgs Japan Kagoshima (Shibushi SFC '87 d
a-sal USA California (Salton Sea) '78 w
a-mk USA Florida (Makay Bay) '80 w
c_xm China Fujian Fish. Res. Inst. '87 d
P-ilo Philippines Panay Island AQD SEAFDEC '84 d
p-le Philippines Panay Island Leganes Stn. '84 d
AQD SEAFDEC
P-Ot Philippines Oton River (Panay Island) '84 w
i-ja Indonesia Java '86 d
s-sin Singapore Natl. Inst. of Aquaculture '86 d
t-son Thailand Sonkia NICA '87 d
t-pu Thailand Puket Marine Inst. '87 d
is-eil Israel Eilat '87 d
J-Sol Japan Shizuoka PE '78 d
J-TKU Japan Univ. Tokyo '78 d
J-OTK Japan Oita (kamiura) SFC '87 d
J-SAP Japan Saga PE '84 d
J-SAP-86 Japan Saga PE '86 d
J-NSU Japan Nagasaki Univ. '69 d
J-NSZ Japan Nagasaki PE '86 d
J-NSC Japan Nagasaki CE '86 d
J-NSGT Japan Nagasaki (Goto Island) SFC '87 d
J-NSGT-II Japan Nagasaki (Goto Island) SFC '87 d
J-KAU Japan Kagoshima Univ. '86 d
F-PA France Palavas-les-Flots '87 d
F-PA-11 France Palavas-les-Flots '87 d
F-PA-III France Palavas-les-Flots '87 d
F-PA-IV France Palavas-les-Flots '87 d

Table 2
Buffer systems used for electrophoresis of enzymes.

Electrode buffer Gel buffer
Abbreviated
name Components pH Components pH References

C-A 0.04 m Citric acid, 6.1 Dilute 50 mL of electrode buffer 6.1 Clayton and Tretiak (1972)
adjust pH up to 6.1 with to I liter (Citric acid, 0.002 M).
N-(3-aminopropyl)-morpholine.
C-A 0.04 m Citric acid, 6.9 Dilute 50 mL of electrode buffer 6.9 Clay and Tretiak (1972)
adjust pH up to 6.1 with to 1 liter (Citric acid, 0.002 M).
N-(3-aminopropyl)-morpholine,
then to 6.9 with NaOH.
C-T 0.04 m Citric acid, 8.0 Dilute 50 mL of electrode buffer 8.0 Clayton and Tretiak (1972)
adjust pH up to 8.0 with to I liter (Citric acid, 0.002 M).
Tris-(hydroxymethyl)-methylamine.

Fu et al.: Genetics of the Rotifer Brachionus plicatilis 17
rotifer can be divided into two groups, and that the strains
A S within the same type display further variation.
206 L
o 277 Allozyme Analysis
lk 2t 2t 24-0 905 28-0 3@0[pN
L Among 18 enzymes tested, 10 enzymes showed clear band-
0.792
0 =1 0. 821
Tr 0 s .ng patterns (Table 3). However, bandings for 3 enzymes
(AK, EST, and IDH) were not genetically interpretable.
a75 0a, 5 The number of alleles of each locus are summarized in
S Table 4. On MDH, 3 isozyme loci were recognized, al-
0.619 L B 0: 662 though no alleles were detected at 2 loci. In Table 5 are
C shown the number of L- and S-type strains and the alleles
0.5 0.6 0 0,8 they posses at each locus. The L and S strains differ con-
L siderably in allele profiles. For instance, at Ldh where 8
0.755 1.095 alleles were observed, 9 of 15 L strains possessed the A
D Erl _U allele whereas none of S strains possessed the A allele. In
E ER;@U" 0-7 d8 0'9 10 .'1 .2 contrast, B allele appeared only in the S strains. There were
L considerable genetic differences between L and S strains
0. 3M S F G 0-f568 for 15 alleles at 6 loci. Allele frequencies for each allele at
T' A 9 loci affecting 7 enzymes were estimated for each strain
er V\1A 0'3 Ot0.5 0.6 0'7 'G in which individuals were considered to be genetically iden-
tical. For MDH, however, three zones of banding patterns
appeared. Although two of those three zones were not in-
Figure 3 terpretable as showing allozyme variation, we regarded
Averages and standard deviations of 5 varieties of measurements allele frequency as one if the strain had the bandings in
considered for differentiating L- and S-type strains.

(A) morphological genetical (B)
80 60 40 20 0 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 18
j-P t-son
t-son P- t
i s-e ij kgko-86
i-j aa-s a I
Po tj -k g k
s-Si ni-j .
J-kgk. - 86 i-kay
tpu J-ehp
C_xm a-mk
P-1 ICX.
i-kay Jotk
j-kgs p-1 e
Jnsz P-i lo
J-ehp - i n
J-0tk t-P
1 @j-,
a-m kI _! ,s z
a-s a Ij-P
J-kgko i s-e I
J-TKU PP A-11
F-PA_RF-PA
F-PA JTKU
J-SAP J-SAP
J-NSZ J-NSZ
J-NSGTK J N SGT-11
J-NSGT J_NSGT
i
J_KAU J-NSC
-Sol i-so I
Iii -SAP.86 J-SAP .861 1
i_0TK J-OTK
F-PA-M J-KAU
F-PA_lV F-PA-IVi
J-NSC F-PA-M
J-NSU J-NSU
PC

E@ 9:1

80 60 2 @0_ 20 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
fusion level genetical distance

Figure 4 of similarities among 34 strains by morphological and genetic snalyses. Abbreviated names by capital and small
letters mean that the strain belongs to L and S type, respectively.

18 NOAA Technical Report NMFS 92

Table 3 Table 5
The different enzyme systems of Brachionus plicatilis screened The number of U and S-type strains for each allele at dif-
with various buffers. x no detectable bandings; A ferent enzyme loci. 9 = Great difference in allele posses-
unclear bandings; 0 find bandings. sion between U and S-type strains.

Buffer Relative mobility L-Type S-Type
Locus Allele (15) (19)
C-A C-A C-T
Enzyme (pH 6. 1) (pH 6.9) (pH 8.0) Ldh X 100 2 0
0 A 81 9 0
aGPD X X X 0 B 69 0 5
SDH X X X C 64 4 4
LDH A 0 A 0 D 47 1 12
HBDH X X X E 39 1 3
MDH A 0 A 0, 0 2
ME X A A 0, 1 0
IDH A A 0 Mdh-I A 100 0 3
6PGD A 0 X B 83 13 16
G6PD X A X C 62 15 9
SOD 0 0 A
AAT A 0 A Mdh-II 0 ? - 0 17
AK 0 0 A Mdh-III 0 ? - 15 6
PGM X 0 A 6Pgd A 100 0 1
EST A 0 A * B 91 10 0
ALP A A A BL 75 1 1
ACP A A A C 68 4 1
LAP X X X CL 55 2 0
GPI A A 0 D (-)39 0 10
DL (-)52 0 1
0 E (-)68 0 13
F (-)75 0 1

Sod 0 A 100 0 6
Table 4 0 B 95 10 0
Isozyme loci and number of observed alleles. 0 C 74 0 17
0 D 49 5 0

Enzyme Locus Aflele Subunit structure Aat-I A 100 11 8
B 75 15 15
LDH Ldh 8 Tetramer Pgm-I A 100 4 6
MDH Mdh-I 3 Dimer B 93 3 2
Mdh-II ? C 89 1 2
Mdh-III ? D 86 7 6
6PGD 6Pgd 9 Dimer E 79 3 9
SOD Sod 4 Dimer F 77 0 1
G 70 0 2
AAT Aat-I 2 Dimer 0 1 0
PGM Pgm-I 8 Monomer Gpi 0 A 100 0 6
GPI Gpi 6 Dimer B 78 0 1
C 67 2 6
0 D 13 0 13
0 E 28 13 0
F 0 7 2

the zone. If not, we decided allele frequency on the zone
as zero. According to Nei's formula (1972), the genetic
distances among the 34 strains were estimated from gene
frequencies including estimated values for MDH. The den- of which consists only of S type strains, and the other two
drograrn expressing similarities among the 34 strains was clusters consist only of L-type strains. Although one of the
also drawn from genetic distances (Fig. 413). The 34 strains two large clusters includes the two types of rotifers, the
can therefore be divided into two major groups. One group classification by the genetic distances also pointed out that
consists only of the strains which had been identified as there are great genetical distances between L and S strains.
S type judging by the anterior spine shape. The other Some of the strains which are genetically identical (e.g.,
cluster can be divided again into 3 smaller clusters, one genetic distance = 0) were collected from neighboring loca-

Fu et al.: Genetics of the Rotifer Brachionus plicatilis 19

tions or from the same hatchery, for instances between the 1981. Polymorphosis in size ofrotifer, Brachionusplicatilis, cultured
two strains of p-ilo and p-le or between J-NSGT and with various feeds. Bull. Natl. Res. Inst. Aquacult. 2:1-10.
J-NSGT-11. However, 'in one instance (c-xm and *-otk), Fukusho, K., and M. Okauchi.
1982. Strain and size of the rotifer, Brachionus plicatilis, being
the samples were geographically unrelated. cultured in southeast asian countries. Bull. Natl. Res. Inst.
Aquacult. 3:107-109.
1983. Sympatry in natural distribution of two strains of a rotifer,
Discussion Brachionusplicatilis. Bull. Natl. Res. Ins. Aquacult. 4:135-138.
1984. Seasonal isolation between two strains of rotifer Brachionus
plicatilis in an eel culture pond. Bull. Jap. Soc. Sci. Fish. 50:909.
For comparison, the two dendrograms are shown in the Hirayama, K.
same frame (Fig. 4). The dendrogram patterns for the two 1987. An Approach from the physiological aspect to the problems
methods are very similar, especially with respect to the in present mass culture technique of the rotifer. In Proceedings
L-type strains. of the 15th (1986) U.S. Japan meeting on aquaculture (Al Sparks,
The results indicate that the rotifer Brachlonus plicatill's ed.). U.S. Dep. Commer., NOAA Tech. Rep.
King, C.E., and Y. Zhao.
can be divided into the two types of genetic constitution. 1987. Coexistence ofrotifer (Brachionusplicatilis) dones in Soda Lake,
The results in this report were drawn from 34 strains, Nevada. Hydrobiologia 147:57-64.
collected mainly from western Japan. In the case of the Nei, M.
L-type, the overseas strains obtained came from only one 1972. Genetic distance between populations. Am. Nat. 106:
283-292.
locality. We are now collecting more strains from all over Serra, M., and M.R. Miracle.
the world in order to make a more unequivocal conclusion. 1983. Biornetric analysis ofBrachionusplicatilisecotypes from Spanish
lagoon. Hydrobiologia 104:279-29L
1985. Enzyme polymorphism in Br"hionusplicatilis populations from
Acknowledgments several Spanish lagoons. Verh. Internat. Limno. 22:2991-2996.
1987. Biometric variation in three strains of Brachionus plicatilis
as a direct response to abiotic variables. Hydrobiologia 147:
The authors wish to express their sincere thanks to 83-90.
H. Kayano, Nagasaki University, for his kind advice on Shaw, C.R., and R. Prasad.
the interpretation of allozyme variation, to K. Fukusho who 1970. Starch gel electrophoresis of enzymes - a compilation of
kindly provided photos of L and S strains, and also to the recipes. Biochem. Genet. 4:297-320.
scientists who kindly sent us live samples of the rotifers. Siciliano, M.J., and C.R. Shaw.
1976. Separation and visualization of enzymes on gels. In
Chromatographic and electrophoretic techniques, 4th ed., Vol.
2:185-209 (1. Smith, ed,). William Heinemann Medical Books
Citations Ltd, London.
Snell, T.W., and K. Carrillo.
Clayton, J.W., and D.N. Tretiak. 1984. Body size variation among strains of the rotifer Brachionus
1972. Amine-citrate buffers for pH control in starch gel electro- plicalihs. Aquaculture 37:359-367.
phoresis. J. Fish. Res. Board Canada 29:1169-1172. Snell, T.W., and B.C. Winkler.
Fukusho, K. 1984. Isozyme analysis of rotifer proteins. Biochem Syst Eco
1983. Present status and problems in culture of the rotifer Br"hionus 12:199-202.
plicatilis for fry production of marine fishes in Japan. Symp. Intl. Suzuki, M.
Acuacultura Coquimbo, Chike-Septiembre:361-373. 1983. Taxonomical study on rotifers cultured for fry production.
Fukusho, K@., and 1, Iwamoto. Zool. Mag. 91:657.
1980. Cyclomorphosis in size of the cultured rotifer, Brachionus 1987. IntraspecificvariabdityofBrachionusplicatilis. Hydrobiologia
plicalilis. Bull. Natl. Res. Inst. Aquacult. 1:29-37. 147:45-48.

Present Status of Genetic Studies on Marine Finfish in Japan

KUNIHIKO FUKUSHO

National Research Institute of Aquaculture
Fisheries Agency
Nakatsuhama, Nansei-cho, Watarai-gun, Mie
516-01 Japan

ABSTRACT

The present paper briefly introduces the status of genetic breeding of marine finfish in Japan.
The domestication of exotic species is also described. Selection is the most successful technique
of genetic breeding for marine finfish, even though limited scientific data and experimental results
have been reported. Selection was conducted on red sea bream, I'agrus major, and Japanese flounder,
Paralichthys olivaceus, and their selected strains were supplied for industrial culture. Experiments
on hybridization leading to heterosis were conducted as well as interspecific, intergeneric, inter-
family, back and reciprocal crosses. Few of these hybrids, however, have been widely used by
the industry, except the P. major x crimson sea bream, Evynnisjaponicus. Chromosome manipu-
lation studies such as triploid production and all female production by gynogenesis have been
conducted since 1984 in Japan. These technologies are strongly expected to be adopted in in-
dustrial culture, even though they are currently experimental. Exotic species or strains of marine
finfish have been introduced to Japan and cultured in recent years. Most marine species are
imported to supply seed where local production is inadequate, not to introduce a new industrial
target species or strain, except the red sea bream. Cryopreservation of sperm is used in most
hybridization studies, induction of gynogeriesis, and triploid production. This technology will
no doubt be adopted in androgenesis and gene bank projects for marine finfish.

Introduction tools for improving fish quality. Therefore, various kinds
of experiments on genetics, including chromosome man-
Mariculture of finfish is well developed in Japan as reflected ipulation have been intensively conducted in recent years
by the total harvest in 1986 of nearly 2 x 105 tons. The for marine finfish, despite the short history of mariculture.
number of cultured species is approximately 30 (Table 1) The objective of the present paper is to provide a brief
(Fukusho 1981). Yellowtail, Seriola quinqueradiala (14.6 x introduction on the status of genetic breeding of marine
104 tons), and red sea brearn (3.4 x 104 tons), are the most finfish, except salmonids in Japan. Introductions leading
important species. The Japanese flounder and coho to domestication are considered part of the study of gene-
salmon, Oncorhynchus kisutch, have shown great promise as tics for the purpose of this review.
cultured species, with their production levels increasing
rapidly in recent years (0.3 x 104 and 1.2 x 104 tons, re-
spectively for 1987 data). All the marine species are cul- The Introduction of Exotic Species
tured in net cages except the Japanese flounder which is
usually raised in land based tanks. Exotic species or strains of marine finfish, such as
At present, the total supply is adequate to satisfy demand yellowtail, red sea bream, rockfish, Sebasticus spp., grouper
for yellowtail and red sea bream, taking into considera- Epinephelus spp., knifejaw Oplegnathus spp., have been in-
tion the total cornsumptiorn plus the carrying capacity of troduced to Japan and cultured in recent years. Marine
the culture ground. Therefore, research presently focuses finfish have been introduced for different reasons than have
on the improvement of fish quality (e.g., high growth rate, freshwater fish. Most marine species are imported to supply
high resistance to pollution, good taste and flavor, and seed where the local production is inadequate, not to intro-
favorable color) based on the requirements of culturists and duce a new industrial target species of strain. An excep-
consumers. Genetic breeding is one of the most effective tion is the case of the red sea bream.

22 NOAA Technical Report NMFS 92

Table I
Marine finfish cultured in Japan (Fukusho 1981).

Family Common name Scientific name Pref. No.'

Salmonidae Cohn salmon Oncorhynchus kisutch I
Mugilidae Grey mullet Mugil cephalus I
Oplegnathidae Japanese striped knifejaw Oplegnathus fasciatus 19
Japanese spotted knifejaw 0. Punclatus 7
Serranidae Sea bass Lateolabraxjaponicus 5
Sea bass L. latus I
Groupers Epinephelus spp. 3
Girellidae Nibbler Girella punctala 11
Sparidae Red sea brearn Pagrus major 25
Porgy Sparus sarba 2
Crimson sea brearn Eoynnisjaponica I I
Porgy Acanthopagrus schlegeli 17
Porgy A. latus 2
Pomadasyidae Grunt Parapristipoma trilinea I
Carrangiclae Yellowtail Seriola quinqueradiata 26
Amberjack S. purpurascens 8
Amberjack S. aureovittata 2
Horse mackerel Trachurusjaponicus 20
Striped jack Longirosirum delicatissimus 9
Scombridae Bluefin tuna Thunnus thynnus 3
Siganiclae Rabbit fish Siganusfuscessens 3
Alutericlae Filefish Stephanolepis cirrhifer 2
Tetradontidae Puffer Fugu rubripes 9
Scorpacnidae Rockfish Sebastiscus marmoratus 3
Rockfish Sebastes inermis
Rockfish S schlegel, I
Bothiclae Flounder Paralichthys olivaceus 10

'Number of prefectures where the species was cultured.

A deep reddish color is highly prized in cultured red sea and response to new environmental conditions should always be
bream by the Japanese consumers. A Korean strain, which considered prior to introduction, as with freshwater fish.
is identical to the Japanese strain taxonomically, shows a
much deeper reddish color than the latter (Harada et al.
1988, Harada et al. 1985, Kumai et al. 1986). No difference Hybridization
has been found between the two strains in electrophoresis
analysis of isozymes. The Korean strain is preferred by fish Experiments on hybridization leading to heterosis have con-
farmers because of its deeper reddish color, even though tributcd to the development of larval rearing techniques for
the Japanese strain is superior to the Korean strain in terms marine finfish (Fujita 1961, 1967; Harada 1974, 1975, 1978).
of growth rate (Kumai et al. 1986). The deep reddish color Interspecific, intergeneric, interfamily, back and reciprocal
is caused by its higher content of carotenoid and astaxan- crosses have also been attempted (Harada 1978).
thine in the skin, which results even under the same rear- Hybridization of marine finfish was initiated on puffers
ing and feeding conditions as the Japanese strain (Kumai (1961-67) of which several species are a high prized luxury food
et al. 1986). Hybridization between the Japanese and despite the fact that parts of these fish are highly toxic (Fujita
Korean strains has been conducted to provide a hybrid with 1967) (Table 2). The Fisheries Laboratory of Kinki University
a deeper reddish color and higher growth rate (Harada has further promoted hybridization to improve fish quality.
et al. 1988). Several successful and promising hybrids were produced that are
Both fertilized eggs and juveniles of Japanese red sea superior in growth rate, survival rate, body color, and meat
bream have been exported to foreign countries; Thus quality to each parent fish (Table 3) (Harada 1974, 1975, 1978).
marine finfish have been introduced world-wide and mari- The "kindai" (Oplegnathusfasciatus x 0. punclatus), which ab-
culture has expanded. In each country, they are being breviates the name of Kinki University and which means sea
cultured as an introduced species. Precise investigation and brearn and sounds like "golden fish" injapanese, is significantly
research of a new marine species' biological characteristics superior in growth rate, survival rate, and handling to each

Fukusho: Genetic Studies on Marine Finfish in Japan 23

Table 2 Table 3
Hybridization of puffers (Fujita 1967). Hybrids of marine finfish produced at the Kinki Univer-
Hatching rate sity (Harada 1974, 1975, 1978).
Female Male (%) Female x Male

Takifugu pardalis x T. poecilonotus 95 Pagrus major x Acanthopagrus schulegeli 1964
T. poecilonolus x T. pardalis 95 P. major x Sparus sarba 67
T rubripes x T. xanthopterus 93 Oplegnathusfasciatus x A. schlegeli 68
Lagocephalus lunaris spadiceus x T. niphobles 0 0. fasciatUS a X 0. Punctatus 69
L. 1. spadiceus x Canthigaster rivulata 0 Seriola quinqueradiata' x S. auroeovittata 70
S. quinqueradiaea X S. purpurascens 70
S. purpurascens' x S. aureovittata 71
S. aureovittata X S. purpurascens 72
P. majora x Evynnis japonicus 73
parent fish (Kumai 1984; Harada et a]. 1986). Because red 0. punialus' x 0. fascialus 73
sea bream. are usually cultured in protected bays with vary- Auxis thazard x Euthynnus affinis 76
ing salinity, tolerance to low salinity is an important char- 'Promising hybrid.
acteristic. Therefore, a useful hybrid of the red sea bream,
and the porgy, Acanthopagrus shulegeli, was developed which'
showed both improved tolerance to lower salinity than the
maternal fish plus faster growth and better taste than the
paternal one (Harada 1975). The Nagasaki Prefectural wild red sea bream caught in the Akashi area in the Seto
Institute of Fisheries has also conducted studies on the Inland Sea. Production of selected strains has also been
hybridization of marine finfish. A hybrid of a sparid, Sparus intensively conducted on the Japanese flounder and the
sarba, and the porgy, A. shulegell', was produced in order Japanese sriped knifej aw, 0. fasnatus, at the Fisheries Lab-
to combine the highest growth characteristics of S. sarba oratory of Kinki University (Harada 1975, 1978).
and the low salinity resistance of A. schulegeli (Kitajima and
Tsukashima 1983). However, the hybrid showed mater-
nal characteristics in both its morphological and physio- Chromosome Manipulation
logical characteristics (Kitajima and Tsukashima 1983).
The same phenomenon occurred in the hybrid of P. maj'or Since 1984, chromosome manipulation studies such as
and the crimson sea bream, Evynnisjaponicus (Arakawa and triploid production, all female production by gynogenesis,
Yoshida 1986, Arakawa et al. 1988). extraction and synthesis of growth hormone, production
Few of these hybrids have been widely used by industry, of cloned fish, and cell fusion have been conducted in
except the hybrid of P. maj'or x E. japonicus. The reason Japan to produce new strains of marine finfish. These
may be due to 1) conservative consumers to whom appear- technologies have been called "Fisheries Biotechnology."
ance is very important (red sea brearn must look like the In 1985, the Ministry of Agriculture, Forestry, and
wild red sea bream. because theJapanese people always eat Fisheries (MAFF) designed and organized a large-scale
the whole and raw fish for sashimi and sushi; 2) limited scientific project on chromosome manipulation titled
attempts to show clearly the difference in quality between "Development of new breeding techniques by induction
the hybrid and parent fish; and 3) too short a period of of gynogenesis in fish." The National Research Institute
marine finfish culture for the industry to evaluate and of Aquaculture was the leading institution to promote the
utilize new strains or hybrids as well as exotic species. project, along with three Universities (Tokyo University
of Fisheries, Hokkaido University, and Kansei Gakuin
University), two Regional Fisheries Laboratories (Nansei
Selection Regional Fisheries Laboratory and Hokkaido Regional
Fisheries Laboratory of the Fisheries Agency-MAFF), and
Selection is the most successful technique of genetic breed- three Prefectural Institutes of Fisheries (Saitama, Hyogo,
ing for marine finfish, even though limited scientific data and Hokkaido). The target species included the Japanese
and experimental results have been reported. Selected flounder, red sea bream, plaices, Verasper moseri, Limanda
strains of red sea bream have been supplied by the Fish- shrenki, L. punctatissima, Platichthys stellatus, and filefish,
eries Laboratory of Kinki University. These selections are Stephanolepis cirrhifer, Thamnaconus modestus, Aluterus monoceros.
highly desired by fish farmers because their growth rates The Fisheries Agency has also organized and initiated a
are approximately 30-40% higher than the wild forms. scientific project addressing fisheries biotechnology. Several
Mass selection has been conducted at the Fisheries Lab- Prefectural fisheries laboratories have also joined this
oratory of Kinki University over several generations from "Local Biotechnological Study Project" where marine

24 NOAA Technical Report NMFS 92

finfish such as the red sea brearn and flounder are being preservation is also used in most hybridization studies,
studied. induction of gynogenesis, and triploid production.
Triploid and gynogenetic diploid induction techniques In marine fish, cryopreservation has been conducted on
which use cold shock have been used to block the second a variety of species, such as: two species of goby, Glosso-
polar body extrusion for red sea brearn (Arakawa et al. gobius olivaceus, Acanthogobw'flavimanus; porgy; mullet, Mugil
1987; Arakawa and Miyahara 1988; Fukusho et al. 1987b), cephalus; mackerel, Scomberjaponicus; bluefin tuna, Thunnus
porgy (Arakawa et al. 1987), and flounder (Tabata, 1988; thynnus; and puffer, Takaifugu nipholbles (Doi et al. 1982;
Tabata and Gorie 1988a, Tabata et al. 1986). The dura- Kurokura 1983). A recent study on the hybridization of
tion of cold shock intervals are as follows: for red sea bream, red sea bream and crimson sea brearn showed positive
15-20 min. duration of OOC, starting 3 min. after insemina- results with high survival rates and increased growth rates
tion; for porgy, 25 min. duration, starting 1.5 min. after observed when sperm preserved for 6 months was used
insemination; and for flounder, 45 min. duration, start- (Kurokura et al. 1986).
ing 3-5 min. after insemination. Suppression of the Ist Thus, gamete preservation is useful for hybridization of
cleavage was achieved by using increased hydrostatic species which spawn in different seasons, genetic breeding
pressure (Tabata and Gorie 1988b). UV irradiation has by chromosome manipulation, transplantation (introduc-
been effective for genetic inactivation of sperm (e.g., tion), and gene bank projects for marine finfish.
1000-2000 erg/mm2 for red sea brearn). Also, sperm of
different species have often been used as an indicator of
the induction of gynogenesis (Fukusho et al. 1987a; Yano Acknowledgments
and Sakai 1988). Triploids were produced in red sea brearn
and porgy (Arakawa et al. 1987, 1988; Fukusho et al. I wish to express my sincere thanks to Dr. F. Brian Davy,
1987b). Thus, various conditions for induction of triploid IDRC, Canada and Dr. Ryo Suzuki, National Res. Inst.
and gynogenesis have been examined for several marine Aquaculture, Japan for their critical reading of this
finfish, and comparisons of growth rate, survival rate, and manuscript.
other biological characteristics have been conducted
through larval rearing (Arakawa and Yoshida 1986; Fuku-
sho et al. 1987a; Tabata et al. 1986; Tabata and Gorie Citations
1988a). Comparison of growth rate during rearing to young
stage and commercial size was also conducted (Tabata and Arakawa, T., and J. Miyahara.
Gorie 1988a), but there is little information to evaluate 1988. Induction of gynogenesis with ultra violet rays in red sea
gynogenetic and triploid fish in view of industrial culture bream, Pagrus major. Bull. Nagasaki Pref. Inst. Fish. 14:37-42.
because the scientific activities have only just started. (In Japanese; English summ.)
Arakawa, T., and Y. Yoshida.
Reports .and papers on the comparison of growth and 1986. Growth, survival and morphologic comparison between fry
maturation rates and sex ratio between chromosome ma- cross bred Pagrus major with Evynnisjaponica and hatchery reared
nipulated fish and common diploid fish are expected to Pagrusu major. Bull. Nagasaki Pref. Inst. Fish. 12:27-35. (In
promote these techniques in industrial mariculture of fin- Japanese; English summ.)
fish. Techniques for examination of ploidy have been Arakawa, T., M. Tanaka, K. Inoue, 1. Takami, and K. Yamashita.
established, but with conflicting result (e.g., appearance 1987. An examination of the conditions for triploid induction by
cold shocck in red sea brearn and black sea bream. Bull. Nagasaki
of males among gynogenetic diploids of the flounder in spite pref. Inst. Fish. 132:25-30. (In Japanese; English summ,)
of a theoretical expectation of all female production) Arakawa, T., C. Kitajima, K. Yamashita, A. Ikeda, and H. limura.
(Tabata 1988; Tabata and Gorie 1988b). These phenom- 1988. Growth and morphology of crossbred Pagrus major with Evynnis
ena could not be explained by the XX and XY sex i.aponica. Bull. Nagasaki Pref. Inst. Fish. 14:31-3 5. (In Japanese;
English summ.)
chromosome theory. Further investigation on embryology Doi, M., T. Hoshino, Y. Taki, and Y, Ogasawara.
and sexual differentiation might be required as well as 1982. Activity of the sperm of the bluefin tuna Thunnus thynnus under
genetic studies. The mechanism of sexual differentiation fresh and preserved conditions. Bull. Japan. Soc. Sci. Fish. 48:
should be clarified to advance the technology of chromo- 495-498. (In Japanese; English summ.)
some manipulation. Fujita, S.
1961. Studies on life history and aquaculture of important puffers
in Japan. Special report of the Nagasaki Pref. Inst. Fisheries,
No. 2, 121 p. (In Japanese)
Cryopreservation of Sperm 1967. Artificial interspecific and intergeneric hybridization among
the tetradontid puffer (Prelim. rep.). Jpn. J. Michurin Biol.,
Experiments on androgenesis have been conducted for 3:5-11. (In Japanese; English summ.)
freshwater fish in Japan. Cryopreservation of sperm in Fukusho, K.
1981. Present status and view of fry production and genetic breed-
combination with androgenesis is useful in preserving en- ing of marine finfish. Fish Genetics and Breeding 6:1-10. (In
dangered species, and also in all male production. Cryo- Japanese.)

Fukusho: Genetic Studies on Marine Finfish in Japan 25

Fukusho, K., M. Okauchi, H. Nanba, M. Hoshi, and H. Tsubaki. Oplegnaihusfasciatus (Temminck and Schlegel). Bull. Fish. Lab.
1987a. Comparison in growth and survival rate among gynogenetic Kinki Univ. 2:1-127. (In Japanese; English summ.)
larvae of red sea bream, being induced by sperm of red sea bream, Kumai, H., M. Nakamura, Y. Kubo, and Asada.
flounder, and striped knifejaw. Proc. MeetingJapan. Soc. Sci. 1986. Comparison of growth and morphological characteristics
Fish. 1987 (Hakodate), p. 149. (In Japanese.) among Japanese, Korean, and Hong Kong red sea bream.
1987b. An attempt of triploid induction of red sea bream, using Proc. Meeting Japan. Soc. Sci. Fish. 1986 (Tokyo), p. 28. (In
fertilized eggs by natural spawning in net cage. Proc. Meeting Japanese.)
Japan. Soc. Sci. 1987 (Hakodate), p. 149. (In Japanese.) Kurokura, H.
Harada, T. 1983. Cryopreservation of fish sperm. Fish Genetics and Breeding
1974. Genetic improvement of sea bream. Yoshoku (Midori- Sci. 8:42-53. (In Japanese.)
shobow), 11:50-54. (Injapanese.) Kurokura H., S. Kasahara, H. Kumai, and M. Nakamura.
1975. Target species of marine finfish for fry production. In 1986. Hybridization of red sea brearn and crimson sea brearn by
Feeding and development of larvae and juvenile (Japan. Soc. Sci. cryopreservation of sperm. Proc. MeetingJapan. Soc. Sci. Fish.
Fish., eds.), p. 90-96. Koseisha-Koseikaku, Tokyo, Japan. (In 1986 (Tokyo), p. 50. (In Japanese.)
Japanese.) Tabata, K.
1978. Cross breeding of marine finfish. Yoshoku (Midori-shobow) 1988. Review: Studies on chromosome manipulation in Hirame,
15:32-35. (In Japanese.) Paralichthys olivaceus. Fish Genetics and Breeding Sci. 13:9-18.
Harada, T., 0. Murata, S. Miyashita, S. Oda, and S. Maeda. (In Japanese.)
1985. Incubation and larval rearing of Korean red sea bream. Tabata, K., and S. Gorie.
Proc. Meeting Japan. Soc. Sci. Fish. 1985 (Tokyo), p. 54. (In 1988a. Comparison of the growth of gynogenetic diploids with con-
Japanese.) trol diploid in Hirame Paralichthys olivaceus in the same tank.
Harada, T., H. Kumi, and 0. Murata. Bull. Japan. Soc. Sci. Fish. 54:1143-1147. (In Japanese; English
1986. Artificial hybrids between Japanese parrot fish and spotted summ.)
parrotfish. Bull. Japan. Soc. Sci. Fish. 52:613-62 1. (In Japanese; 1988b. Induction of gynogenetic diploids in Paralich1hysalivaceus by
English summ.) suppression of the I st cleavage with special reference to their sur-
Harada, T., 0. Murata, and S. Miyashita. vival and growth. Bull. Jpn. Soc. Sci. Fish. 54:1867-1872. (In
1988. Artificial hybridization between Japanese red sea bream and Japanese; English summ.)
Korean red sea bream, and three years culture of the hybrids. Tabata, K., S. Gorie, and N. Taniguchi.
Proc. MeetingJapan. Soc. Sci. Fish., 1988 (Tokyo), p. 276. (In 1986. Verification by isozyme gene marker for gynogenetic
Japanese.) diploidization and triploidization in Hirame, Paralighthys
Kitajima, C., and Y. Tsukashima. olivaceus. Fish Genetics and Breeding Sci. 11:35-41. (In Japa-
1983. Morphology, growth and low temperature and low salinity nese.)
tolerance of sparid hybrids. Jpn. J. Ichthyol. 30:275-283. (In Yano, Y., and Y. Sakai.
Japanese; English summ.) 1988. Introduction of gynogenetic diploids in two species of flat
Kumai, H. fish. Bull. Hokkaido Reg. Fish. Res. Lab. 52:167-172. (In
198C Biological studies on culture of the Japanese parrot fish, Japanese; English summ.)

Recombinant Viral Vaccines in Aquaculture*

JO-ANN C. LEONG, R. BARRIE, H.M. ENGELKING, J. FEYEREISEN-KOENER,
R. GILMORE, J. HARRY, G. KURATH, D.S. MANNING, C.L. MASON,
L. OBERG, and J. WIRKKULA

Department of Microbiology
Oregon State University
Corvallis, Oregon 97331-3804

ABSTRACT

Viral pathogens in aquaculture have largely been controlled by the culling and destruction
of carriers and infected animals and eggs. Because most viral pathogens in aquaculture are trans-
mitted via water and because sensitive animals reside in the neighboring waters, the administra-
tion of attenuated viral vaccines has not been feasible. Attenuated vaccines require costly trials
to assure that these modified live viruses are nonvirulent in all species and that reversion to virulence
does not occur. Killed viral vaccines have been too expensive to produce for the aquaculture
industry. Thus, subunit viral vaccines developed by recombinant DNA techniques are attractive
alternatives for the industry. These vaccines are nonreplicating and inexpensive to produce. The
molecular cloning and expression of viral genes in several host vector systems for the develop-
ment of subunit viral vaccines for aquaculture has been the primary research focus of the authors'
laboratory. Work on the development of such vaccines for infectious hematopoietic necrosis virus
(IHNV), a fish rhabdovirus, and infectious pancreatic necrosis virus (IPNV), a fish birnavirus,
is presented. Laboratory tests of both vaccines in vivo have indicated that fish develop protective
immunity to live virus after vaccination.

Introduction earthen ponds and stable viruses, like the baculoviruses and
the picornaviruses, these disinfection procedures may not
One of the major factors that will have an impact on the work.
success of the aquaculture industry is the control of Another facet of present day controls for viral diseases
diseases. As the industry grows and greater productivity in aquaculture is the requirement for certified pathogen-
demands are madeon facilities, the incidence of disease free stocks and eggs and the use of specific pathogen-free
outbreaks will increase. Thus, the need for more effective water in the facility. When available, these requirements
disease controls has been receiving more attention. The have been very effective in preventing disease outbreaks.
viral diseases are particularly important because there are However, it has not always been possible to obtain disease-
no suitable treatments available. In the United States, there free stocks for highly-prized strains nor economically prac-
are no approved antiviral drugs or vaccines that can be tical to use specific pathogen-free water. Thus, the aqua-
used in the aquaculture industry today. culture industry has a definite need for viral vaccines. Our
At the present time, the control of viral diseases is based group reports here the successful development of two pro-
largely on management. Current recommendations for the totype viral vaccines by recombinant DNA techniques.
control of viral disease outbreak include the destruction of Two viruses were selected for vaccine development
diseased stocks, drainage of ponds, disinfection of con- because these viruses affect economically important aqua-
taminated areas with chlorine, treatment with sunlight or culture species (salmon and trout) in the United States,
lime, and the restocking of the facility with disease-free Europe, and Japan. In addition, these viruses, infectious
stock. These procedures are very expensive and instituted hematopoetic necrosis virus (IHNV), and infectious pan-
with understandable reluctance by the industry. With creatic necrosis virus (IPNV), affect very young fish, and
immunization of large numbers of fish at this size by im-
mersion is fairly easy. IHNV is a rhabdovirus with an
*Oregon Agricultural Experiment Station Technical Paper No. 8961. enveloped virion and glycoprotein peplomers on the

28 NOAA Technical Report NMFS 92

envelope surface; it has a viral genome of single-stranded pared against purified IHNV and IPNV were made as
RNA of negative sense ( 'McAllister and Wagner 1977). In described (Engelking and Leong 1989).
contrast, IPNV is a nonenveloped virus with a single cap-
sid and a genorne of two segments of double-stranded RNA
(Dobos 1975). The techniques that were involved in the Construction of Recombinant Plasmids
construction of recombinant plasmids containing the genes The construction of a recombinant plasmid containing the
for the major immunogenic proteins of these two viruses trp E promoter and the trp E gene fused to an immunogenic
have been described (Kurath and Leong 1985; Huang region of the gene for IHNV glycoprotein gene or the
et al. 1986). The expression of these proteins in bacteria IPNV VP2 gene is shown schematically in Figure 2. The
and the use of these expressed proteins as vaccines will be isolation, cloning and sequence analysis of these genes have
described here. been reported (Kocner et al. 1987; D. S. Manning 1988).
The pATH vectors were the generous gift of A. Tzagaloff
(Dieckmarm and Tzagoloff 1985). The constructions were
Materials and Methods verified by DNA sequence analysis by the dideoxy method
(Sanger et al. 1977). The plasmid pUC 19, which served
Cells, Virus, and Antisera as the negative control for pTA1 in Figure I was obtained
The following viruses here used in this study: the IHNV from Pharmacia, Inc., Piscataway, NJ.
isolate from Round Butte was obtained from W. Groberg
(Oregon Department of Fisheries and Wildlife) and the Immunization Trials in Fish
IPNV isolates, Sp and Buhl, were obtained from R. Hed-
rick, University of California at Davis. The virus used for Bacterial crude lysates were prepared as described (Kleid
challenge studies was prepared by infecting rainbow trout et al. 1981). Proteins were analyzed by sodium dodecyl
(Oncorhynchus mykiss) fry and reisolating the virus from fish sulfate polyacrylamide gel electrophoresis (SDS-PAGE)
dying of IHN disease in the case of the IHNV isolates and and Western immunoblotting as previously described
IPN disease in the case of the IP@NV isolates. Subsequently, (Gilmore et al. 1988). The crude lysates were used to im-
the virus was grown for two passages in chinook salmon munize fish by immersion. Rainbow trout fry at 0.4 g were
embryo cells (CHSE-214 cells) (Fryer et al. 1978). The immunized in sets of 100 fish. Immunization was ac-
tissue-culture supernatant fluid containing the virus was complished by bathing groups of 100 fry in 25 mL of the
used as the challenge virus. The IHNV and IPNV used vaccine preparation (ca. 3 mg/mL total protein concen-
as molecular weight markers in Figure I were prepared tration) for I minute. At that time, the immersion solu-
as described in Kurath and Leong (1985) for IHNV and tion volume was increased to 250 mL with water and fish
Huang et al. (1986) for IPNV. The rabbit antisera pre- were incubated in this diluted solution for an additional

12 LO Figure I
.2 Analysis of bacterial production of trpE-viral gene
I.- tu
E 4 E fusion proteins by antibody reactivity on an elec-
> CL CL
Z
A. 'L
CL B. 1 5 25 1 5 25 1 2 trophorectic transfer blot of a 10 % SDS-polyacr-y-
------- lamide gel of bacterial extracts. (A) Development
of the blot made with anti-IHNV sera. Lane a
110 _VP1 is the trpE-G fusion protein detected in cells con-
-# WWI, taining the recombinant plasmid, p52G; Lane b
66- -4- TrpE-VP2 are proteins detected in cells containing the ex-
45- -VP2 pression vector pATH3, without a viral gene in-
sert; and Lane c is purified IHNV. (B) Develop-
*".mow ment of the blot made with anti-IPNV sera.
31-1 VP3 Lanes a and i contain the molecular weight
VP3A marker proteins: phosphorylase B (110 000 Da);
bovine serum albumin (66000 Da); ovalbumin
(45 000 Da); and carbonic anhydrase (31000 Da).
a b c a b c d e f g h I In lanes b, c, and d are cell lysates from bacterial
Anti-IHNV Sera Anti-IPNV Sera cells containing the plasmid pUC 19 with no viral
insert; the samples were loaded at 1, 5 and 25
- pL respectively in lanes b, c, and d. In lanes c,

Leong et al.: Recombinant Viral Vaccines in Aquaculture 29

2 minutes. These fish were then placed in aquaria of 5 from stained gels and Western blots of total bacterial ex-
gallons with a water flow rate of 0.25 gal/min in a con- tract. In Figure 1, the product of a trpE-IHNV glycopro-
stant water temperature of 10'C. The control fish were tein fusion gene from the plasmid p52G and the major cap-
exposed to saline in the same procedure or left undisturbed. sid protein of IPNV from the plasmid pTAl is shown in
Approximately one month after immunization, the ex- Western blots of the appropriate bacterial lysates. A deter-
perimental and control fish were placed in separate aquaria mination of the DNA sequence of p52G indicated that a
in groups of 25. The fish were exposed to serial log virus 264 bp fragment of the lHNV glycoprotein gene had been
dilutions in I liter of water. The challenge virus was inserted in-frame with the trpE protein to produce a fusion
prepared as described by Engelking and Leong (198 1). In protein of 49000 daltons (49 kDa = 37 kDa [trpEl + 11
Figure 3, the data for fish exposed to 7.2 x 10 + 5 plaque kDa [glycoprotein gene fragment]). In addition, a second
forming units per mL (PFU/mL) is presented. The data fragment of the IHNV glycoprotein gene had been inserted
represents the mean of duplicate experiments. All dead fish out-of-frame adjacent to the 264 bp fragment and this ad-
were assayed for the presence of infectious virus in chinook ditional nucleic acid resulted in I kDa (84 bp extra) more
salmon embryo cells (CHSE-214) as described by Engel- of amino acid residue owing to the fusion protein (Gilmore
king and Leong (1981). et al. 1988).
The IPNV expressing plasmid, pTAl, contained the
entire coding region of the A segment of the viral
Results genome for the isolate Sp. It was constructed so that
Antigen Production in Bacteria the viral genetic information was fused in-frame to the
trpE protein (Figure 2) and all the proteins encoded by the
The size and quantity of virus-specific antigen produced A segment were synthesized in the bacteria. Thus, VP2
in bacteria hosting the recombinant plasmids was estimated (major capsid protein), and VP3 (minor capsid protein)

TrpE

PLASMID WITH EXPRESSION
CLONED VIRAL VECTOR Eco R1
GENE AMP R Bam H1
Hind III
Pst I Pst I

VIRAL GENE

Figure I (Continued)
f, and g are 1, 5 and 25 pL of cell lysates from
bacterial cells containing the plasmid pTA1. Lane VIRAL GENE Restrict with either
h contains purified IPNV. The arrow indicates the Eco Rl
trpE-VP2 fusion protein found in lanes e, f, and g. Restrict with compatible Bam Hl
The symbol VP1 indicates virion protein 1; VP2, enzyme Hind III
virion protein 2; VP3, virion protein 3; and VP3a,
breakdown product of VP3.

TrpE

TrpE
Figure 2 Promoter VIRAL GENE
Construction of the expression vectors for trp E-viral INSERT
gene fusions. The cDNA cloned insert of the IHNV IMMUNOLOGICAL
glycoprotein gene or the A segment of the IPNV 01 SCREENING FOR
genome was restricted with a compatible nuclease EXPRESSION
to permit the insertion of a portion of the viral gene
into the expression vector in the proper reading
frame. The resulting plasmids were used for expres-
sion of a trpE-viral gene fusion protein in E. coli.

30 NOAA Technical Report NMFS 92

100 -

CONTROL
80-
_j

Figure 3
0 60
Immunization of rainbow trout with a subunit vac-
cine against IHNV. Rainbow trout fry (0.4 g) were
Z
Uj 40 immersed in a bacterial lysate (3 mg/mL, contain-
cc ing 10% expressed IrpE-G fusion protein) as de-
Ui
20 VACCINE scribed in Materials and Methods section. The
results are expressed as mean percent total mortality
0 on the ordinate and days after the initiation of viral
0 4 8 12 16 20 24 challenge on the abscissa. There were 25 fish in the
control group and 26 fish in the vaccinated group.
DAYS AF7ER VIRAL CHALLENGE In this particular challenge, the fish received 7.2 x
lo,5 plaque forming units/mL.

50-

CONTROL

40-

Figure 4
30 - Immunization of rainbow trout with a subunit vac-
0 cine against IPNV. Rainbow trout fry (0.3 g) were
20 immersed in a bacterial lysate (3 mg/mL) contain-
Z ing the 1rpE-VP2 fusion protein as described in
Materials and Methods section. The results are
W 10- expressed as percent total mortality on the ordinate
a.
0 0 0 0 VACCINE and days after the initiation of viral challenge on the
0 a 6 0 1 1 1 abscissa. There were 25 fish in both control and vac-
0 10 20 30 40 50 60 cinated groups. The fish received 10" plaque
DAYS AFTER VIRAL CHALLENGE forming units/mL of IPNV-Buhl strain for viral
challenge.

of IPNV-Sp were expressed by this recombinant plasmid Discussion
in bacteria.
We have presented initial findings on the efficacy of bac-
Immunization Trials with Subunit Vaccine terially expressed viral proteins as subunit vaccines for fish.
Both the IHNV and the IPNV vaccines were effective in
Viral challenges provided data on the efficacy of the immunizing fish against lethal viral challenge in laboratory
bacterially expressed protein as vaccine&. A significant trials. Moreover, the vaccinations were carried out on rain-
level of protection (697o) was conferred on fish immu- bow trout fry that were 0.4 g in size. These fish were able
nized with p52G versus unimmunized fish when chal- to respond effectively to the viral vaccine. Previous studies
lenged with the Round Butte isolate of IHNV (Figure 3). of immunization in fish have indicated that the minimum
The glycoprotein used in constructing the fusion pro- size for successful immunization by immersion was 0.8 g
r

tein was derived from this strain. In Figure 4, the pro- for chinook salmon (Fryer et al. 1978) and 1-2.5 g for rain-
tection that was achieved by immunization with pTAI bow trout Uohnson et al. 1982).
against the Buhl isolate of IPNV, a heterologous virus The use of these vaccines with different species of fish
strain, is shown. A decrease in virus-induced mortalities and against a variety of different viral strains must be
from 45 To to 3 To was found for the immunized group of tested. In addition, the duration of effective immunity must
fish. be determined. However, the possibility now exists for

Leong ct al.: Recombinant Viral Vaccines in Aquaculturc 31

developing an inexpensive and effective vaccine for fish necrosis virus. Nucl. Acid Res. 3:1903-1919.
using recombinant DNA technology. Engelking. H.M., and J.C. Leong.
1981. IHNV persistently infects chinook salmon embryo cells.
The development of any vaccine must have safety as well Virol. 109:47-58.
as efficacy as one of its primary considerations. The safety 1989. The glycoprotein of infectious hematopoeitic necrosis virus
of live attenuated vaccines has been questioned for the elicits neutralizing antibody and protective responses. Virus Res.
aquaculture industry because of the nature of the environ- 13:213-230.
ment where the vaccine would be applied. The vaccine has Fryer, J.F., J.S. Nelson, and R.L. Garrison.
to be completely safe for cultured and wild salmonid fish 1978. Immunization of salmonids for control of vibriosis. Mar.
Fish. Rev. 1,0:20-23.
in the watershed. Moreover, the vaccine has to be eco- Gilmore, R.D. Jr., H.M. Engelking, D.S. Manning, and J.C. Leong.
nomical and a subunit vaccine produced in bacteria seems 1988. Expression in Escherichia coli of an epitope of the glycopro-
to be a viable alternative. The initial trials of the subunit tein of infectious hematopoietic necrosis virus protects against viral
vaccines reported here suggest that bacterially expressed challenge. Bio/Technology 6:295-300.
viral proteins, even in crude lysates, can be used as effec- Huang, M., D.S. Manning, M. Warner, E.B. Stephens, andJ.C. Leong.
1986. A physical map of the viral genome for infectious pancreatic
tive and economical viral vaccines. necrosis virus Sp: Analysis of cell-free translation products derived
from viral cDNA clones. J. Virol. 60(3):1002- 1011.
Johnson, K.A., J.K. Flynn, and D.F. Amend.
1982. Onset of immunity in salmonid fry vaccinated by direct im-
Acknowledgments mersion in Vibrio anguillarum and Yersinia ruckeri bacterins. J. Fish.
Dis. 5:197-205.
This publication is the result of research sponsored by Kleid, D.G., D. Yansura, B. Small, D, Dowbenko, D.M. Moore,
Bonneville Power Administration Contract DE-A179- MJ. Grubman, P.D. McKercher, D.O. Morgan, B.H. Robertson,
84BP16479, Project 84-43 (G.R. Bouck and R. Morinaka and H.L. Bachrach.
1981. Cloned viral protein vaccine for foot-and-mouth disease:
served as the Contracting Office Technical Representatives Responses in cattle and swine. Science 214:1125-1129.
on the project) and, in part, by Oregon Sea Grant with Koener, J.F., C.W. Passavant, G. Kurath, and J.C. Leong.
funds from the National Oceanic and Atmospheric Ad- 1987. Nucleotide sequence of a cDNA clone carrying the glycopro-
ministration, Office of Sea Grant, Department of Com- tein gene of infectious pancreatic necrosis virus, a fish rhabdo-
merce, under grant no. NA85AA-D-SGO95 (project no virus. J. Virol. 61(5):1342-1349.
Kurath, G., and J.C. Leong.
R/FSD-l 1) and from appropriations made by the Oregon 1985. Characterization of infectious hernatopoietic necrosis virus
State Legislature. We thank Rebecca Day for typing the in mRNA species reveals a non-virion rhabdovirus protein. J.
manuscript. Virol. 53:162-468.
McAllister, P.E., and R.R. Wagner.
1977. Virion RNA polymerases of two salmonid rhabdoviruses. J.
Virol. 22(3):839-843.
Citations Manning, D.S.
1988. Deletion mapping and Expression of the Large Genomic seg-
Dieckmann, C.L., and A. Tzagoloff. ment of Infectious Pancreatic Necrosis Virus. Doctoral Diss.,
1985. Assembly of the mitochondrial membrane system. J. Biol. Oregon State Univ., Corvallis, Oregon.
Chem. 260:1513-1520. Sanger, F., S. Nicklen, and A.R. Coulson.
Dobos, P. 1977. DNA sequencing with chain-terminating inhibitors. Proc.
1975. Size and structure of the genome of infectious pancreatic Natl. Acad. Sci. USA 7,t(12):5463-5467.

Genetic Monitoring of Pacific Salmon Hatcheries

ROBIN S. WAPLES, GARY A. WINANS, FRED M. UTTER,
and CONRAD MAHNKEN

Northwest Fisheries Center
National Marine Fisheries Service, NOAA
2725 Montlake Blvd. East
Seattle, WA 98112

ABSTRACT

In the last few decades, and in response to substantial reductions in the abundance of wild
populations of Pacific salmon, an enormous amount of resources in both Asia and North America
has been devoted to artificial propagation programs. Several factors increase the possibility of
rapid (often detrimental) genetic change in cultured populations, but genetic considerations are
often overlooked in the effort to increase short-term productivity. Here, we discuss recent studies
using electrophoretic data for chinook salmon, Oncorhynchus tshawytscha, that address three im-
portant concerns for hatchery populations: levels of genetic variability, stability of allele frequen-
cies, and genetic interactions (due to straying or overplanting) between hatchery and wild popula-
tions. Results indicate that although there is no evidence for a general reduction in levels of genetic
variability in hatchery stocks relative to wild populations from the same geographic area, allele
frequencies over a period of one generation changed much more in samples from hatchery popula-
tions in Oregon than in nearby wild populations. The genetic changes in the hatchery stocks
appear to be due to a combination of two factors: genetic drift due to reduced effective popula-
tion size, and (in some cases) the infusion of genes from other populations through straying or
transfer of broodstock between hatcheries.

Introduction Materials and Methods

As a consequence of increased fishing pressure, loss of The electrophoretic data discussed here were collected over
spawning habitat, and blockage of migratory routes, re- the last decade at the National Marine Fisheries Service
turns of wild anadromous salmonids in the Pacific North- laboratory in Seattle. A considerable database exists for
west have declined substantially in this century. In part all the North American species of Pacific salmon, Oncorhyn-
to mitigate these losses, an extensive public hatchery sys- chus, but here we consider only data for chinook salmon,
tem has been developed during the last several decades. 0. tshawy1scha; for this species, data are available for pop-
Throughout most of this period, management practices at ulations from California to Alaska. Whole juvenile fish or
the hatcheries have been dictated primarily by production tissue samples (muscle, liver, eye, heart) from adult fish
demands, and relatively little consideration has been given were collected in the field and stored at - 70'C until
to the genetic quality of released fish and their effects on analyzed. Starch gel electrophoresis was performed as
wild fish. The availability of large amounts of data pro- described by Aebersold et al. (1987). Each sample was
duced by protein electrophoresis over the last decade has surveyed for genetic variation at up to 100 presumptive
made possible a critical evaluation of the genetic status of gene loci, and genotypes inferred from the phenotypic
Pacific coast hatchery populations of salmonids. Here, we banding patterns (see Utter et al. 1987 for discussion) were
summarize results from several recent studies which are used to compute allele frequencies and a variety of stan-
pertinent to three important concerns: 1) levels of genetic dard indices of genetic variability and differentiation.
variability found in hatchery and wild populations; 2)
stability of allele frequencies in hatchery and wild popula- Levels of Genetic Variability
tions; and 3) genetic interactions (due to straying or over-
planting) between hatchery and wild populations. Recent policy statements (e.g., Northwest Power Planning

34 NOAA Technical Report NMFS 92

Council, 1987) regarding anadromous salmonids express One drawback to the above analysis is that average
two major concerns: that existing levels of genetic diver- heterozygosity is not very sensitive to the presence or
sity be maintained, and that unique gene pools be pre- absence of uncommon alleles. Although they contribute
served. Loss of genetic variability is a real concern for little to the measurement of heterozygosity, such alleles are
managed populations because constraints on money, space, potentially very important to a population because they
and other resources often limit the size of the breeding allow a greater degree of plasticity in response to changes
population. In a closed population, approximately 1/2N, in the environment. The presence of numerous allctes (even
of the existing genetic variation is lost each generation, with those at low frequency) in a population ensures that each
N, being the effective number of breeders (Crow and generation, many genotypic combinations are produced
Kimura 1970). The effective population size (N,) is less upon which natural selection might act. Because alleles at
than the actual number (N) if the sex ratio is uneven or low frequency are easily lost if the effective breeding size
if the variance in reproductive success among families is is small, the average number of alleles per locus is a more
large-both factors that might be influenced by hatchery sensitive indicator than average heterozygosity of undesir-
management procedures. Furthermore, if population size able changes in the genetic makeup of a population.
changes over time, long-term N, is determined primarily According to Utter et al. (1989), the average number of
by the effective number of breeders in the generation(s) alleles per locus for seven hatchery and six wild popula-
with smaller size. Therefore, a population bottleneck (re- tions from Oregon were similar (1.74 and 1.68, respec-
duced effective breeding size in one or a few generations) tively). This lends additional support to our conclusion that
can contribute appreciably to the long-term erosion of the wholesale reduction of genetic variability reported in
genetic variability. some hatchery populations of Salmo (e.g., Stahl 1983) has
To determine whether these effects are important in apparently not occurred in chinook salmon hatcheries in
Pacific salmon, we examined two measures of genetic the Pacific Northwest.
variability (average heterozygosity and effective number This result is encouraging, but by no means constitutes
of alleles per locus) in a series of hatchery and wild pop- a clean bill of health for hatchery populations. If the genetic
ulations of chinook salmon. The occurrence of consis- makeup of the source populations is to be perpetuated as
tently lower levels of genetic variability in hatchery stocks accurately as possible, it is important not only to conserve
would suggest that artificial propagation has caused pop- overall levels of genetic variability, but also to avoid large
ulation bottlenecks. The heterozygosity data, however, changes in frequency of the alleles present. For example,
provide no evidence of the erosion of genetic variability consider a locus with two alleles (A and a), sampled in a
in cultured populations of chinook salmon in the Pacific population at two times, with the following frequencies
Northwest. In each case where data are available for a observed-time 1: A = 0.8, a = 0.2; time 2: A = 0.2,
comparison (Fig. 1), hatchery and wild populations from a = 0.8. Hardy-Weinberg expected heterozygosity (2Aa
the same area have very similar levels of heterozygosity. = 0.32) remains unchanged, but allele frequencies have
This result differs from that reported in a number of studies shifted drastically. Clearly, it is also important to monitor
of Atlantic salmon, Salmo salar, and rainbow, Oncorhynchus allele frequencies over time in artificially propagated
mykiss, cutthroat, 0. clarkil, and brown trouts, Salmo trutta populations.
(review, Allendorf and Ryman 1987); some cultured
populations of these species have been found to have greatly
reduced levels of heterozygosity relative to the ancestral Temporal Stability
wild stocks. of Allele Frequencies
Some interesting trends are apparent in the heterozy-
gosity data for chinook salmon but these relate to geo- To evaluate the temporal stability of allele frequencies, we
graphic differences rather than to differences between examined electrophoretic data for 21 coastal chinook
hatchery and wild populations. In the Columbia River salmon populations from Oregon and California that were
basin, coastal populations have higher heterozygosity than sampled in each of two years (Waples and Teel 1990). For
do lower river populations, which in turn retain more each population, allele frequencies in the two samples
genetic variability than Snake River populations from far- were compared at an average of 10 polymorphic loci. For
ther upstream (Fig. 1). Populations from the Klamath and each locus, a contingency chi-square test was used to test
upper Fraser rivers also show reduced levels of genetic the hypothesis that the population frequencies were un-
variability relative to those closer to the coast (Georgia changed. Results of these tests are very revealing (Table
Strait, Puget Sound). Presumably, these differences reflect 1). For the three California hatchery and the nine Oregon
the essentially independent evolutionary histories of the dif- wild populations, the number of single locus tests show-
ferent areas and, perhaps, the smaller population size or ing a significant change in allele frequency (1/16 = 6%
increased frequency of population bottlenecks in the up- to 7/88 = 8 To) was close to the value (517o) expected to
river populations (Winans 1989). arise from sampling error, while the figure for the nine

Waples et al.: Genetic Monitoring of Pacific Salmon Hatcheries 35

155 Upper
5 10 Fraser
River
15 Georgia 5
10 Strait
0
5

- - - - - - - - - - - - - T
15 8 t I
j Puge
Sound

0 L

8
15

10
Oregon - 7 -6 5 Lower 1 3
Columbia
::1.
Coast 10 Snake
XX
0 River River
10
5
5
0
0
3 2
15 Klamath
10- River

5-
0 - - - - - -

155
acramento
10 Ri;ver
5 i@i:@ Wild
i Hatchery
0

CL
0 Figure I
Heterozygosity in hatchery Heterozygosity Comparison of average heterozygosity values for
and wild Chinook salmon hatchery and wild stocks of chinook salmon by
geographic area.

Oregon hatchery populations was much higher (29/81 cients to explain such large frequency shifts in a single
36% of all tests showing significant allele frequency generation. Waples and Teel also showed that the observed
change). In addition, combined chi-square tests over all differences can be accounted for by genetic drift only if the
loci indicate very significant (P<0.01) or highly significarit effective number of breeders in the Oregon hatcheries
(P<0.001) changes in allele frequency between 1981 and averaged about 50 or less. Examination of brood stock data
1985 samples in eight of the nine Oregon hatcheries (Table indicates that effective population size may indeed have
1). been quite low in at least some of the hatcheries (Waples
Possible causes of short-term allele frequency change in- and Teel 1990).
dude natural selection, genetic drift, and migration. In the Another possibility is that some of the genetic changes
present example, selection appears to be an unlikely cause, resulted from the infusion of new genes during the transfer
given Waples and Teel's (1990) demonstration that it is of fish between hatcheries, or from natural straying into
necessary to invoke unrealistically large selection coeffi- the hatcheries. Evidence to support this hypothesis comes

36 NOAA Technical Report NMFS 92

Table 1
Twenty-one chinook salmon populations sampled in each of two years. Number of significant
(a = 0.05) single locus chi square tests comparing allele frequencies in two years are shown, and
significance levels are given for combined chi-square test over all loci and a test for gametic
disequilibria (n.s. not significant).

Between-year cmparisons' Gametic disequilibriad

Sample size Single locus
Population' 1981 1985' (no. sig./total) All loci 1981 1985,

Oregon wild populations
Alsea 94 50 0/11 n.s. ii.s. n.s.
Chetco 100 93 1/7 n.s. n.s. n.s.
Coquille 115 50 1/12 n.s. ii.s. n. s.
Nehalern 141 50 1/9 n. s. n.s. n. s.
Nestuca 60 50 1/9 n.s. n.s. n.s.
Siletz 92 50 0/11 n.s. n.s. n.s.
Sixes 100 50 1/8 0.01 n.s. n.s.
Siuslaw 82 34 1/11 n.s. n. s. n.s.
Tillamook 88 50 1/10 n.s. n.s. n.s.
No. sig./total 7/87 1/9 0/9 0/9
Oregon hatchery populations
Cedar Creek 99 100 4/9 0.001 0.01 0.001
Cole R. (S) 113 50 1/9 n.s. n.s. n.s.
Cole R. 50 100 5/13 0.01 n. s. n. s.
Elk R. 100 100 2/9 0.001 n. s. n. s.
Fall Creek 100 100 2/7 0.01 0.001 n. s.
Rock Creek (S) 100 100 4/9 0.001 0.05 0.001
Salmon 99 100 5/8 0.001 n. s. 0.001
Trask (S) 100 100 3/10 0.001 n. s. 0.05
Trask 100 100 3/7 0.01 0.01 0.001
No. sig./total 29/81 8/9 4/9 5/9
California hatchery populations
Iron Gate 99 50 1/8 n. s. n. s. n. s.
Trinity (S) 50 100 0/5 n.s. 0.05
Trinity 100 50 0/3 n.s. n.s. n.s.
No. sig./total 1/16 0/3 0/2 1/3

'Spring run denoted by (S); all others are fall run stocks.
'Samples taken in 1983 for Oregon wild populations, 1984 for California populations.
'Data from Waples and Teel (in press).
'Data from Waples and Smouse (1990).

from gametic disequilibrium analysis, a powerful means reasons. First, the transfer of eggs, fry, and brood stock
of detecting samples which are actually a mixture of among hatcheries is a common occurrence that complicates
distinct gene pools. Gametic disequilibrium (the non- the problem of identifying the genetic makeup of hatchery
random association of alleles at different gene loci) populations. Second, strays of hatchery or transplanted fish
occurs as the result of a mixture of gene pools that dif- may have an adverse effect on wild populations adapted
fer in allele frequency at two or more loci (Nei and Li to local conditions. The genetic consequences of such
1973). admixtures are difficult to evaluate by traditional methods
(physical tags, behavioral observations) because the pres-
ence of exotic fish in a population does not ensure that they
Genetic Interactions Between will interbreed with the residents and produce viable off-
Hatchery and Wild Populations spring. If the potential source populations can be identified
and adequate genetic markers are available, estimates of
Admixtures (mixtures of fish from more than a single gene the mixture fractions are possible (Campton 1987). How-
pool) involving hatchery populations are a concern for two ever, in many cases the populations possibly contributing

Waples et al.: Genetic Monitoring of Pacific Salmon Hatcheries 37

to a mixture are unknown or cannot be characterized Citations
genetically. For such a "blind" mixture, gametic dise-
quilibrium analysis is a potentially powerful tool for evalu- Aebersold, P.B., G.A. Winans, DJ. Teel, G.B. Milner, and F.M. Utter.
ating the null hypothesis that the sample could have come 1987. Manual for starch gel electrophoresis: A method for the detec-
from a single gene pool. tion of genetic variation. Dep. Commer., NOAA Tech. Rep.
61:1-19.
To evaluate the possibility of genetic admixture in the Allendorf, F.W., and N. Ryman.
above example, we used a multilocus analysis of gametic 1987. Genetic management of hatchery stocks. In Population
disequilibrium (Waples and Smouse 1990) that considers genetics and fishery management (N. Ryman and F. Utter, eds.),
data for all pairs of loci simultaneously. No unusual levels p. 141-159. Univ. Washington Press, Seattle.
of multilocus gametic disequilibrium were found in the Campton, D.E.
1987. Natural hybridization and introgression in fishes. In Popula-
California hatchery or Oregon wild populations (I of 23 tion genetics and fishery management (N. Ryman and F. Utter,
samples with significant disequilibria at a = 0.05 level; eds.), p. 161-192. Univ. Washington Press, Seattle.
Waples and Smouse 1990), but the situation was quite Crow, J.F., and M. Kimura.
different in the samples from the Oregon hatcheries (9 of 1970. An introduction to population genetics theory. Harper and
Row, New York, 591 p.
18 tests significant; Table 1). These findings are consis- Nei, M., and W-H. Li.
tent with the hypothesis that, in addition to genetic drift, 1973. Linkage disequilibriurn in subdivided populations. Genetics
a mixture of gene pools contributed to the changes in allele 75:213-219.
frequency observed in some of the Oregon hatchery Northwest Power Planning Council.
populations. 1987. Columbia River Basin Fish and Wildlife Program. Portland,
Gametic disequilibrium analysis has considerable poten- Oregon, 246 p.
Stahl, G.
tial for assessing the extent of genetic interactions among 1983. Differences in the amount and distribution of genetic varia-
hatcheries and between hatchery and wild populations. tion betw .een natural populations and hatchery stocks of Atlantic
It may help in the identification of wild gene pools that salmon. Aquaculture 33:23-32.
have been relatively unaffected by genes from hatchery Utter, F., P. Acbersold, and G. Winans.
1987. Interpreting genetic variation detected by electrophoresis.
populations and therefore merit conservation efforts. In In Population genetics and fishery management (N. Ryman and
other cases, where the objective is to supplement and F. Utter, eds.), p. 21-45. Univ. Washington Press, Seattle.
enhance wild production, gametic disequilibrium analysis Utter, F., G. Milner, G. Stahl, and D. Teel.
.provides a means of monitoring the effectiveness of 1989. Genetic population structure of chinook salmon in the Pacific
transplants from the hatcheries. The possibilities for both Northwest. Fish. Bull., U.S. 87:239-263.
types of genetic interactions, as well as the necessity for Waples, R.S., and P.E. Smouse.
1990. Genetic disequilibriurn analysis as a means of identifying mix-
monitoring them, are likely to increase in the near future. tures of salmon populations. American Fish. Soc. Symp.
Large scale supplementation of wild populations and 7:439-458. (In press.)
expansion of hatchery production are planned to achieve Waples, R.S., and DJ. Teel.
the goal of the current Columbia River Basin Fish and 1990. Conservation genetics of Pacific salmon. 1: Temporal changes
in allele frequency. Conserv. Biol., Vol. 4:144-156.
Wildlife Program (Northwest Power Planning Council Winans, G.A.
1987)-doubling the run size of anadromous salmonids in 1989. Low levels of genetic variability in spring-run chinook salmon
the Columbia River basin. of the Snake River. N. Am. J. Fish. Management 9:47-52.

Successful Gene Transfer in Fish

S.J. Y00N,1 Z. LIU,3 A.R. KAPUSCINSKI,2 P.B. HACKETT,3
A. FARAS,4 and K.S. GUISEI
Departments of Animal Science,' Fisheries and Wildlife,2
Genetics and Cell Biology,' and Microbiology4
University Of Minnesota
1988 Fitch Avenue
St. Paul, MN 55108

ABSTRACT

The neo gene, which confers resistance to the neomycin analog drug G418, has been success-
fully transferred into newly-fertilized dechorionated goldfish eggs by microinjection. Multiple
copies of the gene were incorporated into the high molecular weight fraction of fish DNA (i.e.,
the genomic DNA). RNA dot blots indicate specific neo mRNA synthesis. Gene transfer efforts
using mammalian-derived growth hormone genes, a marker gene for P-galactosidase, and a variety
of promoters are discussed. Current work which also includes the isolation of piscine promoters
and genes for peptide hormones is mentioned.

Introduction the United States, four primary groups are known to be
using microinjection of various plasmid constructs to pro-
Novel genes were first introduced into mice in 1979 by duce transgenic fish. These laboratories are, in addition
Gordon et al. (1980). The technology did not receive wide to our group at the University of Minnesota, those of Drs.
attention until Palmiter et al. (1982) transferred a rat Powers and Chen at Johns Hopkins, Dunham at Auburn,
growth hormone gene linked to a mouse metallothionein and Ellinger and Kohler at Southern Illinois. All four U.S.
promoter, creating a line of mice that grew significantly groups are pursuing similar goals: production of transgenic
faster and ultimately larger than control mice. This series fish by transfer of growth-hormone gene constructs using
of experiments captured the imagination of a wide variety microinjection as the primary transfer technique. Our
of researchers seeking to improve economic traits in domes- group reports here the successful transfer of a marker gene,
tic and semidomestic animals. While gene transfer ex- neo, into goldfish via microinjection, as a step toward the
perimentation is currently progressing in most species of goal of transfer of economically important genes.
economic importance, including mammalian farm animals
and poultry, no group of organisms shows more promise
for the dramatic interaction of transferred genes than fish. Materials andMethods
For over three decades it has been known that fish are quite Egg Preparations
responsive to injections of crude and purified growth hor-
mone (Pickford and Thompson 1948; Adelman 1977). Suc- Spontaneous ovulation of goldfish was accomplished by the
cessful transfer and expression of growth-hormone genes methods of Stacy et al. (1979). Sexually mature fish were
in fish are thus expected to produce a similar, dramatic kept under a long photoperiod (16 h light, 8 h dark). On
response. day one, breeding fish were transferred from stock aquaria
Since 1984, multiple groups worldwide have been pur- to standing-water breeding aquaria at 13 � 1 'C. Aquaria
suing the goal of producing transgenic fish. Laboratories were supplied with floating artificial plants, and the water
in Japan, England, France, and the People's Republic of temperature was increased to 21 � 1'C overnight and kept
China have published results of these attempts in medaka, at that temperature through spawning. Spontaneous ovula-
Oryzias sp. (Ozato et al. 1986), rainbow trout (Oncorhyn- tion usually occurred during the last half of the dark phase
chus mykiss) (Maclean and Talwar 1984; Chourrout et al. on day 3. If no ovulation occurred on day 3, the fish were
1986), goldfish (Carassius auratus) (Zhu et al. 1985), and injected intraperitoneally with 3 mg/kg body weight of carp
loach (Misgurnus anguillicaudatus) (Zhu et al. 1986). Within pituitary gland extract. About 10 hours later, the fish were

40 NOAA Technical Report NMFS 92

artificially spawned and eggs and sperm were collected Tris - HCI pH 8. 0, 0.1 M EDTA, 0. 5 % SDS buffer with
separately. 100 yg/mL proteinase K on ice using a Dounce homo-
Eggs were fertilized by the milt after they were mixed genizer. High molecular weight DNA was extracted as per
with well water in an open petri dish in batches of approx- Maniatis et at. (1982).
imately 100 eggs. Ten minutes after fertilization, the eggs DNA dot blot analysis was used to detect the neo gene.
were dechorionated by a six minute incubation in 0.217o Approximately 5 Mg of genomic DNA were denatured in
trypsin (Zhu et al. 1985). Dechorionation was stopped by 0.4 M NaOH, and following addition of an equal volume
treatment with 5 % fetal bovine serum in Holtfreter's solu- of 2 M ammonium acetate, were spotted onto nitrocellu-
tion (Grand et al. 1941). Dechorionated eggs were washed lose using a Hybri-dot system. The filter was baked in vacuo
several times in Holtfreter's solution and transferred to for 2 hours at 800C. The filter was prehybridized in 5 x
Holtfreter's solution on a charcoal-agar petri dish, which SCC (I x SSC = 0. 15 M NaCl, 15 mM Na citrate), 5 x
provided a dark background for microinjection. Denhardt's solution (1 x Denhardt's = 0.02% Ficoll,
0.02% bovine serum albumin, 0.02% polyvinylpyrroll-
Microinjection done), 50 mM sodium pyrophosphate pH 6.5, 0.1 % SDS,
0. 1 mg/mL denatured calf thymus DNA, and 0. 1 mg/mL
Plasmids were prepared by amplification, lysis in SDS yeast tRNA at 650C for at least 4 hours. The filter was
(sodium dodecylsulfate), and CSC12 centrifugation (Mani- hybridized with I x 106 cpm of 32p labeled probe in the
atis et al. 1982). Plasmids were linearized with restriction same solution used for prehybriclization for 20 hours at
endonuclease Kpnl, extracted with phenol/chloroform, 65'C. The filter was washed three times in 2 x SSC, 0. 1 %
ethanol precipitated, and redissolved in 88 mM NaCl, 10 SDS at room temperature for 10 minutes, and three times
mM Tris - Hcl pH 7.6 to a final concentration of 25 ng/;AL. in 0. 1 x SSC, 0. 1 17o SDS at 600C for 30 minutes, dried,
Borosilicate glass needles with an inner tip diameter of ap- and exposed to Kodak XAR5 x-ray film at - 80'C (Mani-
proximately 2 Vrn were filled with plasmid solution. Micro- atus et al. 1982).
injection was performed with a Brinkman MM33 micro- Southern blot analysis was performed by complete diges-
manipulator with the injection volume controlled by the tion of 10,ug of genomic DNA with Kpnl, BaniHI, SstI,
timing of the insertion/withdrawal interval and constant or Mlu I followed by electrophoresis on a 0. 8 Olo agarose gel.
fluid flow. It is estimated that 2 nanoliters of soluti'on con- Transfer to nitrocellulose was made following the method
taining I X 106 copies of neo gene are delivered. DNA was of Southern (1975). The Southern blot was probed using
released into the center of the germinal disc prior to the the same hybridization conditions as the dot blot analysis.
first cleavage. Microinjected eggs were allowed to develop
in Holtfreter's solution until the blastula stage, and re- RNA Analysis
turned to well water after the blastula stage. Mock injec-
tions were performed as controls, using only the buffer Anterior halves of frozen 90'C) neo DNA positive fish
component of the injection solution. plus several control and injected fish were homogenized
by Polytron in a solution containing 4 M guanidinium
Plasmid Construction isothiocyanate, 0. 1 M P-mercaptoethanol, 0. 5 % sarkosyl,
5 mM sodium citrate, pH 7.0. Total RNA was isolated
The primary construct used, pRSVneo, contains the neo as per Maniatis (1982). RNA dot blot analysis was used
gene, whose product, aminoglycoside 3'-phosphotrans- to test for neo gene expression. Total RNA was dissolved
ferase, confers resistance to the neomycin analog G418, in 6 x SSC, and 7.417o (v/v) formaldehyde, and heat de-
and a Rous sarcoma virus (RSV) promoter. A BamHl- natured at 65'C for 15 minutes. Serial amounts of RNA
HindIll restriction fragment containing the neo gene and (0.25, 0.5, 1, and 2 Mg) were directly dotted onto nitro-
mart of SV40 intron was isolated from pSV2neo (Ameri- cellulose equilibrated with 6 x SSC. The RNA was fixed
can Type Culture Collection) and ligated into the poly- to the filter by baking under vacuum at 80'C for 2 hours.
linker of pUC 119. A 3 40-bp fragment containing the RSV The prehybridization was carried out for more than 4 hours
promoter/enhancer region was ligated to the regenerated at 60'C in 50 To deionized formamide, 5 x SSC, 5 x Den-
HindIII site. This construct is termed pRSVneo in this hardt's solution, 50 mM sodium phosphate, pH 6.5, 0.1 Olo
paper. Plasmid construction was by standard methods SDS, 100 lAg/mL denatured calf thymus DNA, and 100
(Maniatis et al. 1982). Mg/mL yeast tRNA. The filter was then hybridized with
2 x 106 cpm/mL of 32P-labeled probe in the same buffer
DNA Analysis for 20 hours at 601C. After hybridization, the filter was
washed three times with 2 x SSC, 0. 1 % SDS at room
Genomic DNA was isolated from posterior halves of 1 to temperature for 10 minutes, three times with 0.2 x SSC,
2 month-old (approx. 8 g) fish stored at - 900C. Fish were 0. 2 To SDS at 68'C for 30 minutes, and exposed to Kodak
individually homogenized in three volumes of 0.2 M XAR5 x-ray film at - 801C.

Yoon et al.: Successful Gene Transfer in Fish 41

A B C D E F G H I J K L

2
3
4
5
6
7
8

Figure 1
DNA dot blot analysis of goldfish microinjected with pRSVneo. Total genomic DNA probed with
330 bp HindIII-BarnHI restriction fragment of the promoter region of the neo gene. Rows 7 and 8
present dilutions of pRSVneo with (row 8) or without (row 7) added genomic DNA. Dilution series
is A-B, 0.5 copies; C-D, I copy; E-F, 5 copies; G.H, 25 copies; and K-L, 100 copies per genome.
Test fish DNA was spotted twice on filter producing series 1-5, A-F and an identical replicate 1-5,
G-L. Dots row 1A-IF (replicate IG-IL) are of six noninjected control goldfish. Dots row 2A-2F
and 3A-3E (replicate 2G-2L and 3G-3K) are of pRSVneo-injected goldfish DNA. Dots 3B and 3D
indicate presence of approximately five copies of neo gene. Dots 3F, 4A-4F, and 5A-5C (replicate 3L,
4G-4L, and 5G-5I) are of pRSVneo-injected fish subjected to G-418 drug selection (4 mg/mL), ar-
ranged in increasing survival time from 20 hours (3F) to greater than 8 days (5C). Dots 5D-5F are
test dots for detection of possible neo-hybridizable sequences in pink salmon DNA.

Probe Labeling tained above 60% saturation by periodic aeration. The
Plasmid restriction fragments were separated by I 01o aga- temperature was maintained at 20 � I'C. The pRSVneo-
rose gel electrophoresis in TAE buffer (0. 04 M Tris-acetate I injected or non-injected control goldfish at a population
2mM EDTA pH 8.0). Excised bands were phenol:chloro- density of 12 g/L were exposed to 4 mg/mL solutions of
form extracted, and the DNA was precipitated by ethanol G-418. Mortality of fish was monitored every four hours
and sodium acetate and redissolved in TE (10 mM for the first 24 hours and every 12 hours afterward. Dead
Tris - HCI pH 7.6, 1 mM EDTA). Fragments were labeled individuals were removed upon observation and stored in
by the procedure of Feinberg and Volgelstein (1983) to liquid nitrogen. Drug selection was continued until 507o
0.5-2 x 109 cpm/Mg. DNA probes were used without puri- of the control fish in 4 mg/mL G-418 had died. Surviving
fication from unincorporated nucleotides. fish were transferred into fresh well water upon termina-
The 2.3-kb fragment including the neo gene was also sub- tion of the test.
cloned into pTZ 18R under the control of the T7 promoter
to produce a transcript complementary to neo mRNA. One Results
Mg of linear DNA was transcribed using the method of
Schenborn and Mierendorf (1985) to produce probes used The survival rate for microinjected fish has ranged from
in RNA dot blot analysis. 10% in the early experiments to nearly 507o at the cur-
rent time. The results reported here are from early injec-
G-418 Selection tion studies, where, despite the high mortality rates, we
were still able to effect the transfer and apparent expres-
For the drug selection screenings, solutions of G-418 in sion of the transferred gene.
static water aquaria were aerated to achieve initial satura- Figure 1 shows the results of DNA dot blot analysis of
tion levels of dissolved oxygen. Dissolved oxygen was main- mock- and pRSVneo-injected goldfish. None of the mock-

42 NOAA Technical Report NMFS 92

P Kpo Bam Sst Big
P P9 1 2 3 1 2 3 1 2 3 1 2 3

Figure 2
Southern analysis of genomic DNA from three goldfish microinjected with pRSVneo. Fish #1 was negative by DNA dot
blot analysis, while fish #2 and #3 represent fish from dots 3B and 3D of Figure 1. Lanes p and pg represent linearized pRSVneo
and pRSVneo plus genomic DNA, respectively.

injected fish showed positive hybridizaiton to the neo probe, (Fig. 1), was confirmed in the RNA dot blot of Figure 3.
while two of the pRSVneo-injected fish showed approx- Here, one out of six fish tested showed strong expression
imately five copies of the gene per genome. Figure 1 also of neo RNA when probed with a complementary neo RNA
shows a series of DNA dot blots for pRSVneo-injected fish probe. In order to rule out the possibility of DNA con-
that were subjected to drug selection with G-418. The fish tamination of the RNA used in *he RNA dot blot, we
were grouped by order of the time of death during drug treated the RNA from the positive fish of Figure 3 with
selection. One fish, which survived the full course of selec- DNase and RNase in a separate experiment (Fig. 4).
tion, but died on day 8 of fungal infection, showed an ap- RNase destroyed the hybridization to the neo RNA probe,
parent hybridization signal of about I copy per genome. while DNase had little effect. This indicates little if any
Pink salmon DNA was included as control DNA. Note that DNA contamination of the RNA.
only the pink salmon DNA dot blots of Figure 1 (Dots
5D-5F), show high background hybridization to the probe
in both control and injected fish. Unlike goldfish, pink Conclusion and Discussion
salmon apparently contain genomic sequences that are
closely related to neo. The results reported here showed successful transfer of a
Figure 2 shows results of a Southern blot analysis of marker gene neo that encodes resistance to the drug G-418.
genomic DNA cleaved with four different restriction en- Stable integration of the transferred gene was suggested
zymes and probed with the 330-bp neo probe. The neo probe but could not be conclusively proven at this time. Expres-
detected both linearized plasmid and a higher molecular sion of the marker gene was confirmed through detection
weight band in the BamHI and Sstl digests. This may in- of neo mRNA in the transgenic fish. Functional resistance
dicate a concatenated integration of the estimated five conferred by pRSVneo was suggested by the survival
copies of the gene, but conclusive evidence of integration through drug selection of a putative transgenic fish con-
will require breeding studies. taining an extimated single copy of the neo gene.
Expression of the transferred neo gene which was sug- These results indicate that microinjection is a viable
gested in the resistance of a fish containing neo sequences method for transfer of selectable genes into fish. Micro-

Yoon et al.: Successful Gene Transfer in Fish 43

.25 .5 1, 2
A ONase RNase

+ +

C
D 5 RNA

Figure 4
Figure 3 Total RNA from fish E of Figure 3 probed with
RNA dot blot analysis of total RNA from six RNA complementary to neo RNA transcript. RNA
pRSVneo- injected fish, spotted at 0.25, 0.5, 1 was spotted before ( - ) and after ( + ) treatment with
and 2 ;Ag RNA per dot for each fish. RNase-free DNase or DNase-free RNase.

injection is, however, currently a labor- and patience- bovine growth hormone constructs into fish eggs, as well
intensive technique, and promises to remain so for the near as injecting a second marker gene construct using P-galac-
future. Microinjection works well upon dechorionated eggs tosidase. While bovine growth hormone gene constructs
where needle placement can be monitored, but is less may produce growth enhancement in transgenic fish, our
effective in blind injections. Therefore, we are keenly in- intent is to effect growth enhancement by transfer of
terested in development of a selectable marker for use in species-specific growth hormone genes. To this end, we are
either blind injection or in mass transfer techniques. Our isolating piscine growth hormone genes and fish gene pro-
group has been exploring the use of mass transfer tech- moters so that the ultimate transgenic, growth-enhanced
niques to overcome the tedium of injection of the substan- fish will contain no DNA sequence from outside its own
tial numbers of eggs necessary to generate the optimal species. We feel that this will simplify regulatory approval
transgenic fish. While mass transfer efficiencies may be of the transgenic fish for human consumption, and enable
orders of magnitude less than those achieved by microin- the transferred genes to interact with the natural regula-
jection, the ability to manipulate tens of thousands of eggs tory mechanisms in the fish in a more normal fashion.
at a time should compensate for the loss in transfer effi-
ciency. The key element necessary to make mass transfer
a reality is the development of a selection system to differ- Acknowledgments
entiate transgenic from nontransgenic fish. The selectable
marker neo meets this criterion, and should prove effective The work outlined here was supported in part by Minne-
in the selectionof transgenic fish when used in co-transfer sota Sea Grant R/3 to K.S. Guise, by a grant from the
schemes with other genes. Legislative Commission on Minnesota Resources of the
A variety of mass transfer techniques are under inves- State of Minnesota to A. Faras, K.S. Guise, A.R. Kapu-
tigation by our group as well as by other laboratories in scinski, and P.B. Hackett, and by the Minnesota Experi-
the United States. These range from techniques based on ment Station (K.S. Guise).
the CaP04 procedures used in gene transfer in tissue
culture, to binding of plasmid DNA to sperm and effect-
ing transfer at the time of fertilization, to electroporation. Citations
Studies proceeding along these lines should soon yield
general mass transfer protocols that will be successful in Adelman, 1. R.
a broad spectrum of fish species. 1977. Effect of bovine growth hormone on growth of carp (Cyprinus
The establishment of the neo gene as a selectable marker carpio) and the influences of temperature and photoperiod. J. Fish
Res. Board Can. 34:509-515.
is a step in the generation of transgenic fish with enhanced Chourrout, D., R. Guyomard, and L.M. Houdebine.
growth hormone expression. We are currently injecting 1986. High efficiency gene transfer in rainbow trout (Salmo gaird-

44 NOAA Technical Report NMFS 92

neri Rich.) by microinjection into egg cytoplasm. Aquaculture Palmiter, R.D., R.L. Brinster, R.E. Hammer, M.E. Trumbauer,
51:1443-1450. M.G. Rosenfeld, N.C. Birnberg, and R.M. Evans.
Feinberg, A.P., and B. Vogelstein. 1982. Dramatic growth of mice that develop from eggs microin-
1983. A technique for radiolabelling DNA restriction enclonuclease jected with metallothionein-growth hormone fusion genes. Nature
fragments to high specific activity. Anal. Biochem. 132: 300:611-615.
6-13. Pickford, G.E., and E.F. Thompson.
Gordon, J.W., G.A. Scangos, D.J. Plotkin, J.A. Barbosa, and 1948. The effects of purified mammalian growth hormone on the
F. H. Ruddle. killifish Fundulus heteroclitis (Linn). J. Exp. Zool. 109:367.
1980. Genetic transformation of mouse embryos by microin- Schenborn, E.T., and R.C. Mierendorf Jr.
jection of purified DNA. Proc. Nad. Acad. Sci. U.S.A. 77: 1985. A novel transcription property of Sp6 and T7 RNA poly-
7380-7384. merases: Dependence on template structure. Nucleic Acids Res.
Grand, C.G., M. Gordon, and G. Cameron. 13:6223-6236.
1941. Neoplasma studies: Cell types in tissue culture offish melan- Southern, E.M.
otic tumors compared with mammalian melanomas. Cancer Res. 1975. Detection of specific sequences among DNA fragments
1:660-666. separated by gel electrophoresis. J. Mol. Biol. 98:503-517.
Maclean, N., and S. Talwar. Stacy, N.E., A.F. Cook, and R.E. Peter.
1984. Injection of cloned genes into rainbow trout eggs. J. Em- 1979. Spontaneous and gonadotrophin- induced ovulation in the
bryol. Exp. Morphol. 82:187. goldfish, Carassius auratus L. Effects of external factors. J. Fish.
Maniatis, T., E.F. Fritsch, and J. Sambrook. Biol. 15:349-361.
1982. Molecular cloning: A laboratory manual. Cold Spring Har- Zhu, Z., G. Liu, L. He, and S. Chen.
bor Laboratory, Cold Spring Harbor, New York. 1985. Novel gene transfer into the fertilized eggs of goldfish
Ozato, k., H. Kandoh, H. Inohara, T. Iwamatsu, Y. Wakamatsu, and (Carassius auratus L. 1758). Z. Angew. Ichthyol. 1:31-34.
T. S. Koada. Zhu, A., K. Xu, G. Li, Y. Xie, and L. He.
1986. Production of transgenic fish: Introduction and expression 1986. Biological effects of human growth hormone gene micro-
of chicken crystallin gene in medaka embryos. Cell Differ. 19: injected into the fertilized eggs of loach Misgurnus anguillicaudatus
237-244. (Cantor). Kexue Tangbao 31:988-990.

Clonal Ginbuna Crucian Carp as a Model for the
Study of Fish Immunology and Genetics

TERUYUKI NAKANISHI and HIROSHI ONOZATO

National Research Institute of Aquaculture
Tomaki
Mie 519-04, Japan

ABSTRACT

The lack of suitable inbred lines of fish for immunological study in which lymphocytes and
blood samples could be collected have led to the use of clonal ginbuna crucian carp, Carassius
gibelio langsdorfii. Distributed widely in Japan they include both bisexual diploid and tetraploid
types and a unisexual (all female) triploid type which reproduces gynogenetically by omitting
Mciosis I. The kinetics of immunity transfer by immune leukocytes was first examined using
isogeneic crucian carp. Splenic cells were most effective in transferring immune reactivity, followed
by pronephric, then mesonephric cells. However, antibody titres were very low or absent when
the recipients received transferred thymic cells. Antibody productivity was most successfully con-
ferred when cells were taken from 7-day postimmunized donors. The level of antibody titre in
recipients reached its peak on day 7 for mesonephric cells and on day 14 for pronephric and splenic
cells. Transferability of immune leukocytes was also compared in isogeneic, allogeneic and
xenogeneic crucian carp to examine their major histocompatibility complex (MHC) regulation
in adoptive immunity. Adoptive transfer by pronephric cells was successful in isogeneic and weak
histocompatibility (H) gene-disparate transfer systems, while antibody productivity was not
transferable in xenogeneic and strong H gene-disparate transfer systems. In allogeneic fish,
however, antibody productivity was transferable by the transfer of cells in some recipients that
rejected the allograft in an acute fashion.

Introduction present review, we describe our recent studies on the fish
immune system using clonal crucian carp and discuss the
Most of the information on the cellular immune mechan- excellent potential of this fish as a model for studying fish
ism in mammals has been obtained by in vivo and in vitro immunology and genetics.
experiments using histocompatible animals. In fish, how-
ever, suitable inbred lines for immunological study are not
available. Although inbred strains of small teleosts have Clonal State of Triploid Ginbuna
been established by sibmating (e.g., platyfish and sword-
tails, Xiphophorus spp., (Kallman 1964); guppy, Lebistes Naturally occurring gynogenetic populations are included
reticulatus, (Schroder and Holzberg 1972); and medaka, in two cyprinid species: silver crucian carp, Carassius auratus
Oryzias latipes, (Taguchi 1980), they are too small to col- gibeho (Cherfas 1966), and ginbuna, C. gibeho langsdo!fii(for-
lect lymphocytes and blood samples. Furthermore, gene- mally C. auratus langsdorfil, Kobayashi 19 7 1). Ginbuna have
tically identical animals are not currently available for the widest distribution in Japan, and include both bisex-
practical use, even though chromosomal set manipulation ual diploid (2n = 100) and tetraploid (4n = 206) types, and
techniques have seen great improvement in recent years unisexual (all female) triploid (3n = 156) types (Kobayashi
and homozygous clonal fish have been obtained by the sup- et al. 1970, 1977). In unisexual ginbuna, which reproduce
pression of meitosis I and successive prevention of Mciosis gynogenetically, the first polar body formation is skipped
Il (e.g., zebra fish, Brachydanio rerio, (Streisinger et al. as the result of lacking Meiosis I (Kobayashi 1976). There-
1981), medaka (Naruse et al. 1985) and rainbow trout, On- fore, the progenies of these fish belong to a clone having
corhynchus mykiss, (Onozato 1990). the same genotype as the mother, as evidenced by scale
Fortunately, clonal crucian carp-naturally occurring grafting and electrophoreses (Onozato 1981; Murayama
gynogenetic fish-are distributed widely in Japan. In the et al. 1984). Onozato (1981) has shown that the electro-

46 NOAA Technical Report NMFS 92

phoretic pattern of isozymes LDH, MDH, a-GPD and the Table I
muscle protein of all progenies coincide with that of their Antibody response to erythrocyte antigens in crucian carp.
mother. Of most interest is the fact that ginbuna was found Fish were given three injections at 2-day intervals with 5
to be a heterozygous clone from isozyme analyses (Ono- uL/g body weight of 20% erythrocyte intraperitoneally.
zato 1983).
Natural populations of triploid ginbuna consist of several Antigens Individual HA@ titres 1092) Mean � SID
clones which differ in strong or weak histocompatibility Bovine 5, 5, 4, 1, 0 3.0 � 2.1
genes (H-genes). Allografted scales reciprocally exchanged Horse 12, 10.5, 10, 9.5, 9.5 10.3 � 0.9
among strong H gene-disparate members were rejected Sheep 7, 7, 7, 7, 6.5 6.9 � 0.2
within one week in an acute fashion, whereas allografts Guinea-pig 10, 9.5, 8, 7, 5 7.9 � 1.8
among weak H gene-disparate members remained intact Chicken 10, 9.5, 9, 9, 8 9.1 � 0.7
until 4 to 5 weeks before rejection (Nakanishi and Ono- 'Hemagglutination.
zato 1988).

Ginbuna Immunology and Genetics and adjusted to appropriate concentration. All animals
Variability in Immune Reactivity were given an intraperitoneal injection (5 pL/g body
weight) every 3 days for 6 days. Immunized fish were bled
It is a well-known fact that immune responses are gene- from the caudal blood vessels with an heparinized syringe,
tically controlled. Our greatest concern was to determine and the plasma samples were analyzed for hemagglutina-
to what extent the variation exists in response to erythro- tion titres by a microtiter method. Fish weighing 20-30
cyte antigens in genetically identical animals. If the vari- g were maintained in tanks with running water at 25 � I OC
ation is not great, these animals could be used in various and fed pellet. Anesthetization was performed with ethyl
studies such as the examination of vaccines and drugs. As p-amino benzoate (100 mg/L) prior to manipulation. Horse
well as using fewer animals, we could also expect the results red blood cells (HRBC) were found to be the best antigen
to be more reproducible compared to those obtained using for inducing high titres of antibody in crucian carp (Table
outbred animals. 1). We also examined the effect of concentration on the
We examined several erythrocyte antigens for monitor- diversity in immune responses using outbred nigorobuna
ing immune reactivity. Erythrocytes in Alsever's solution crucian carp, Carassius carassius grandoculis. At a high con-
were purchased from Teikoku Zoki Ltd (Tokyo), washed centration of 2017o, most of fish elicited similar titres. At
three times with phosphate-buffered saline (PBS, pH 7.2) lower concentrations, however, greater variation was ob-

Table 2
Effect of dosage HRBC (horse red blood cells) antigen on the antibody response of out-
bred nigorobuna crucian carp at 25'C.

Concentration
of antigen Weeks' Individual HAb titres log2) Mean C.V.1

20% 2 11, 10, 10, 8.5, 7 9.3 0.15
3 11, 10.5, 10, 8, 8 9.5 0.13
557o 2 7.5, 7, 6, 6, 0 5.3 0.51
3 9, 7, 7, 7, 2.5 6.5 0.33
10/0 2 9, 7, 5, 5, 0 5.2 0.58
3 9.5, 7, 7, 6.5, 0 6.0 0.53
0.10/0 2 7, 5.5, 4, 4, 3 4.7 0.30
3 8, 7.5, 6.5,6.5, 5 6.7 0.15
0.01% 2 1, 1, 0, 0,0 0.4 1.22
3 4.5, 3, 2.5, 2, 1 2.6 0.45
0.0010/0 2 0, 0, 0, 0, 0 0.0 0.00
3 6, 4, 3, 0, 0 2.6 0.90

'Interval between the first injection and the bleeding.
'Hernagglutination.
'Coefficient of variance.

Nakanishi and Onozato: Immunology and Genetics of Crucian Carp 47

Table 3
Comparison of variability in antibody responses to HRBC between allogeneic and isogeneic
crucian carp.

No. of
animals Individual HA' titres 1092 ) b Mean C.V.1

20% Allogeneic 15 13(l), 12(l), 11(2), 10.5(l), 10(4) 10.0 0.14
9.5(3), 9(l), 8.5(l), 7(l)
Isogencic 15 13(3), 12.5(l), 12(3), 11.5(2) 11.5 0.10
11(3), 10.5(l), 10(l), 9(l)
1 % Allogeneic 15 9(l), 8(l), 7.5(l), 7(5), 6(3) 6.4 0.22
5.5(l), 4.5(2), 3.5(l)
Isogeneic 15 B(l), 7.5(l), 7(5), 6(6), 5(i), 4(l) 6.4 0.15
0.017o Allogeneic 10 7(2), 5(2), 4(l), 3.5(2), 3(l), 4(2) 4.2 0.41
Isogeneic 10 10(1), 8(l), 7(2), 5(l), 4.5(1) 5.7 0.38
4(3), 3(l)

'Hemmaglutination.
'Number of fish is shown in parenthesis.
'Coefficient of variance.

served; some fish showed high titres, and others did not Fish source and the procedures of immunization, leuko-
respond at all (Table 2). There is no apparent difference cyte collection, and cell transfer have been described in
between allogeneic and isogeneic crucian carp even at lower Nakanishi (1987a). In brief, donors were injected intra-
concentrations (Table 3). That is, genetically identical gin- peritoneally at 2-day intervals with 5 IiL HRBC/g body
buna showed similar considerable variation to that of the weight. Donors used for day-3 transfer were given a single
outbred nigorobuna in immune responsiveness. injection, those for day-5 transfer were given two injec-
Similar results have been obtained from the study of tions and those for day-7 or after were given three injec-
genetic influence on the diversity in growth rate using tions. Recipient fish were subsequently infused with lym-
clonal crucian carp (Nakanishi and Onozato 1987). In this phoid cells at a rate of 0. 1 mL of 5 x 107 cells/mL per
study a great variation in size and weight was found, even fish. Pronephric and mesonephric cells from a fish were
in clonal crucian carp, indicating the occurrence of a injected into two or three fish and thymic and splenic cells
superior group when the fry were reared in small tanks at were transferred into one fish. Recipients were evaluated
high,densities with food of different partical size. by measuring the serum antibody titre using standard
These results suggest that immune reactivity or growth methods of hernagglutination in microtiter plates.
is not only dependent on genetic state but is greatly influ- In a comparison of transferability between lymphoid
enced by other factors which may have been formed in organs, splenic cells were most effective in transferring
correlation with environmental circumstances. That is, immune reactivity, followed by pronephric, then meso-
individuals have met with different microenvironments nephric cells (Fig. 1). Little or no antibody titres were
throughout the life span, including egg state in the ovary, elicited when the recipients received transferred thymic cells
and the differences in physiological conditions must have (Figs. 1, 2). The optimal period for collecting the cells from
occurred among individuals later on. the donor was determined by examining the antibody titres
in the plasma of the recipients which received cells taken
from immunized donors on days 3, 5, 7, 10, 14, and 2 1.
Kinetics of Transfer of Antibody productivity was most successfully conferred
Immunity by Immune Leukocytes when cells were taken from immune donors 7 days after
immunization (Fig. 1). In recipients, antibodies were de-
Adoptive transfer of immunity can be successful only when tected within one day after transfer and the titre reached
recipients show no alloimmune response. However, the peaked levels on day 7 for mesonephric cells and day 14
histocompatibility system of fish has been found to be well for phronephric and splenic cells (Fig. 2). These results
developed (Kallman 1970; Borysenkc, 1976), with rejec- agree well with the kinetics of plaque-forming cells (PFCs
tion often occurring in an acute fashion even among sib- = antibody-producing cells) and circulating antibodies
lings. Therefore, adoptive transfer is only possible when described by Nakanishi (1987a). That is, PFCs were
inbred or clonal fish are used. detected on day 3 and peak response was observed 5 to

48 NOAA Technical Report NMFS 92

7
6 /0\ %
0 5
T
0
4 0
<
o
3
ID

0 Figure I
Kinetics of leukocyte transfer. Pronephros mesonephros
(0), spleen (A) and thymus (A) cells taken from immunized
3 7, 10 14 21 donors on day 3, 5, 7, 10, 14 and 21 are injected into recip-
Day of Cell Transfer ients. Each point represents mean hernagglutination titre in
the plasma for 5 to 10 recipients.

6 -
0
'L 0-
5 -
0 0--
<
:E 4-
C
0

3- 0
Figure 2
2- Antibody production kinetics. Recipients receiving
7-day postimmunized leukocytes from pronephros ( 0
<2
mesonephros (0), spleen (A), and thymus (A) are tested
1 3 5 7 14 21 for antibody levels in the plasma. Each point represents
Days After Cell Transfer mean hernagglutination titre in the plasma for 5 to 10
recipients.

7 days after the first injection. On the other hand, cir- between the PFC response and transferability for individual
culating antibodies were detected 5 days after the first fish was observed in other lymphoid tissues (Nakanishi
injection and reached their peak on day 14. In the pres- 1987a). This lack of correlation between the PFC response
ent study transfer of thymic cells was not successful, and transferability might be attributed to the existence of
even though the number of PFCs of the thymus was sim- different developmental stages or of heterogeneous popula-
ilar to that of the pronephros and the spleen. This findin tions of antibody-producing cells and the necessity of cell
g leads to the idea either that thymic PFCs are not the cells collaboration. In any event this basic infornation. obtained
involved in the transfer of immunity or that thymic cells by using isogeneic animals can be useful for further in-
need collaboration with other sensitized lymphoid cells to vestigations of cellular immune mechanisms on fish im-
produce antibodies. In addition, no direct relationship mune systems.

Nakanishi and Onozato: Immunology and Genetics of Crucian Carp 49

Transferability of Immune Pronephric B I to K1 clones, which differed in strong H gene as evi-
Cells in Isogeneic, Allogeneic and denced by the acute rejection of allografts. Little or no
Xenogeneic Transfer Systems antibody titres were detected in the recipients 7 days after
transfer (Table 4). On the other hand, interclonal transfer
The major histocompatibility complex (MHC) of mam- between weak H gene-disparate clones was successful.
mals and birds consists of numerous genes involved in acute Antibody titre of recipients 7 days after transfer was similar
allograft rejection, cell collaboration, and cytotoxic lym- to that of intraclonal (isogeneic) transfer (Table 4), although
phocyte effective functions (Klein 1977). In lower verte- recipients rejected allografted scales from donors 1-2
brates, a single genetic region homologous to the MHC months after grafting. Transferability of immune pro-
has been described in the anuran amphibian (Du Pasquier nephric cells in xenogeneic transfer systems was also ex-
et al. 1975; Kaufman et al. 1985; Nakamura et al. 1986). amined. Pronephric cell transfer was carried out from
The teleost is the lowest vertebrate in which a MHC is immunized kinbuna to unimmunized ginbuna. No anti-
suspected to exist, owing to their vigorous rejection of body was found and all xenografted scales from kinbuna
foreign tissue grafts, though knowledge of a MHC in fish were rej .ected within 5 days in ginbuna (Table 4). These
is sparse. The present study was undertaken to examine results suggest that transferability of immune cells is strictly
the transferability of antibody reactivity by immune pro- controlled by "MHC."
nephric cells among isogeneic, and xenogeneic crucian carp The most interesting results were obtained when allo-
in order to analyze the correlation between transplanta- geneic transfer using pronephric cells was perforned among
tion antigens and determinants involved in cell collabora- siblings of kinbuna that had rejected allografts of each other
tion or cell-mediated lympholysis. in an acute manner. Two experiments were performed,
In these experiments we used two clones of ginbuna (B 1, each with different times of cell collection from the immu-
D3) from Okushiri Island in the vicinity of Hokkaido Island nized donors: 7 days (Fig. 3A) and 14 days (Fig. 3B) after
and one clone (Kl) from Lake Kasumigaura, lbaragi immunization. In each study, one of five recipients showed
prefecture. In addition, siblings of bisexual diploid kin- elevated antibody titre which peaked 2 and 3 weeks after
buna, C. carasslus buergeri (formerly C. auratus susp.), were transfer from the 7-day (Fig. 3A) and 14-day (Fig. 3B)
also used in hopes that some of them might be genetically postimmunized donors, respectively. Furthermore, one of
related. Fish source and histocompatibility relationship the recipients with cells transferred from the 7 days post-
among clones have been previously described (Nakanishi immunization donor showed a fairly high level of antibody
1987b, Nakanishi and Onozato 1988). Cell transfer was titre 5 weeks after transfer, while no antibody was detected
carried out according to the methods described above. Scale one week and three weeks after transfer. These results sug-
transplantation techniques followed Nakanishi (1987c). gest that the locus of "class IP antigens involved in cell
Interclonal transfer between strong H gene-disparate collaboration or cytotoxic lymphocyte effecter function is
clones was carried out to determine transferability of im- limited in polymorphism in comparison to the transplanta-
mune pronephric cells. Transfer of antibody reactivity was tion antigens (class I antigens), because antibody reactiv-
not successful when pronephric cells were transferred from ity was successfully conferred in some donor-host combina-

Table 4
Transferability of immune pronephric cells in isogencic and xenogeneic crucian carp.

Donor-host Weeks after HA' titre of recipients
relationship transfer log')

Isogeneic BI - BI 1 7.5 6 8 7 5 5
transfer 2 6 6 8 9 6 -

Xenogeneic kinbuna - Bl 1 0 0 0 0 1 0
transfer 2 0 0 0 0 0 2

Inter-clonal BI Kl 1 0 1 0 2 1 0
transfer V 2 0 0 0 0 0 0

Inter-clonal Bi D3 1 9 6 7.5 4 4 J,
transfer 11' 2 8.5 5.5 7.3 4 3 +

'Hemagglutination.
'Cells were transferred from 131 clone to a KI clone that differs in strong H gene.
'Cells were transferred from Bl clone to a D3 clone that differs in weak H gene.

50 NOAA Technical Report NMFS 92

mice and rats. Indeed, clonal ginbuna are a useful tool from
6 many aspects of fish immunology as shown in this review.
There are of course some limitations for using ginbuna;
one is the difficulty of crossing because of their gynogene-
tic reproduction and the other is their heterozygous state.
Even so, ginbuna will continue to be one of the best models
4 for the study of fish immunology until the establishment
of inbred or clonal fish by means of chromosomal manipu-
3 lation technique.

2
Citations
.2 -12
I Borysenko, M.
W
1976. Phylogeny of immunity: an overview. Immunogenetics 3:
305-326.
7 Cerfas, N.B.
1966. Natural triploidy in the female of the unisexual variety of
Z 6 the silver crucian carp (C. auratus gibeho Bloch). Genetika. 5:
< 16-24.
W
X 5 Du Pasquier, L., X. Chardonnens, and V.C. Miggiano.
1975. A major hostocompatibility complex in the toad Xenopus laevis
(Daudin). Immunogenetics 1:482-494.
4 Kallman, K.D.
1964. An estimate of the number of histocompatibility loci in the
3 teleost Xiphophorus maculatus. Genetics 50:583-595.
1970. Genetics of tissue transplantation in Teleostei. Transplant.
2 Proc. 2:263-271.
Kaufman, J.F., M.F. Flajnik, L. Du Pasquier, and P. Riegert.
<2 1985. Xenopus MHC class Il molecules. I: Identification and struc-
tural characterization. J. Immunol. 13+:3248-3257.
Klein, J.
0 1 2 3 4 5 6 1977. Evolution and function of the major histocompatibility sys-
WEEKS tem: facts and speculations. In The major histocompatibility
system in man and animals (D. Gotze, ed.), p. 339-378.
Springer-Verlag, Berlin, New York.
Figure 3 Kobayashi, H.
(A) Transferability of immune pronephric cells from donors 7-days 1971. A cytological study on gynogenesis of the triploid ginbuna
postimmunization in allogencic kinbuna crucian carp. Cells (Carassius auratus langsdo@kz). Zool. Mag. 80:316-322.
(5 x 10') per fish were intravascularly transferred and then anti- 1976. A cytological study on the maturation division in the oogenic
body titres in the plasma of recipients were determined every week process of the triploid ginbuna (Carassius auratus langsdorfu). Jpn.
after transfer. Each line represents the antibody titre of individual J. Ichthyol. 22:234-240.
crucian carp. (B) Transferability of immune pronephric cells from Kobayashi, H., Y. Kawashima, and N. Takeuchi.
donors 14 days post-immunization in allogeneic kinbuna crucian 1970. Comparative chromosome studies in the genus Carassius,
especially with a finder of polyploicly in the ginbuna (C. auralus
carp. langsdorfit). Jpn. J. Ichthyol. 17:153-160.
Kobayashi, H., N. Nakano, and M. Nakamura.
1977. On the hybrids, 4n ginbuna (Carassius auratus langsdorfii) x
kinbuna (C. auratus-subsp.), and their chromosomes. Bull. Jpn.
Soc. Sci. Fish. 43:31-37.
tions even when the recipients rejected an allografted Murayama, Y., M. Hijikata, T. Nomura, and T. Kajishima.
donor's scales in an acute manner. 1984. Analyses of histocompatibility and isozyme variations in a
triploidfish, Carauius auratus langsdo!fii. J. Fac. Sci., Shinshu Univ.
19:9-25.
Summarizing Comment Nakamura, T., A. Sekizawa, T. Fujii, and C. Katagiri.
1986. Cosegregation of the polymorphic C4 with the MHC in the
frog, Xenopus laevis. Immunogenetics 23:181-186.
/ / rI-

Many developments have contributed to the understand- Nakanishi, T.
ing of the immune system in fish over the last decade. 1987a. Kinetics of transfer of immunity by immune leukocytes and
However, more progress cannot be expected without using PFC response to HRBC in isogeneic ginbuna crucian carp,
isogeneic or genetically defined models. Genetically iden- Carassius auratus langsdorfii. J. Fish Biol. 30:723-729.
1987b. Transferability of immune plasma and pronephric cells in
tical animals offer many additional opportunities for ana- isogeneic, allogeneic and xenogeneic transfer systems in crucian
lyzing the immune system of fish, as demonstrated with carp. Dev. Comp. Immunol. 11:521-528.

Nakanishi and Onozato: Immunology and Genetics of Crucian Carp 51

1987c. Histocompatibility analyses in tetraploids induced from 1983. A story of clonal crucian carp. Tansuigyo 9:33-41. (in
clonal triploid crucian carp and in gynogenetic diploid goldfish. Japanese.)
J. Fish Biol. 31:35-40. 1990. Production of clones of homozygous rainbow trout, Oncorhyn-
Nakanishi, T., and H. Onozato. chus mykiss by gynogenesis. (Abstr.) Annu. Meeting jpn. Soc.
1987. Variability in the growth of isogeneic crucian carp, Carassius Sci. Fish. (In Japanese.)
gibdio langsdorfii. Nippon Suisan Gakkaishi 53:2099-2104. Schroder, J.H., and S. Holzberg.
1988. Histocompatibility analyses in a gynogenetic ginbuna cru- 1972. Population genetics of Lebistes reticulatus Peters (poeciliidae;
cian carp, Carassiusgibeho langsdorfii collected from Okushiri Island. Pisces). 1: Effects of radiation induced mutation on the segrega-
(Abstr.) Annu. MeetingJap. Soc. Ichthyol. (injapanese.) tion ratio in postirradiation F2. Genetics 70:621-630.
Naruse, K_ K. Ijiri, A. Shima, and N. Egami. Streisinger, G., C. Walker, N. Dower, D. Knauber, and F. Singer.
1985. The production of cloned fish in the medaka (Oryzias latipes). 1981. Production of clones of homozygous diploid zebra fish
J. Exp. Zool. 236-335-341. Brachydanio relio. Nature 291:293-296.
Onozato, H. Taguchi, Y.
1981. Gynogenesis in fishes. Fish Genetics and Breeding Science 1980. Establishment of inbred strains of the teleost, Oryzias latipes.
6:11-18. (in Japanese.) Zool. Mag. 89:283-301.

Aquaculture of Striped Bass, Morone saxatilis,
and Its Hybrids in North America"'

THEODORE I.J. SMITH

South Carolina Wildlife and Marine Resources Department
P.O. Box 12559
Charleston, SC 29412

ABSTRACT

Increased interest has focused on the culture of striped bass, Morone saxatilis, and its hybrids
(especially M. chrysops) as food fish with the recent decline in fishery landings of striped bass.
Hatchery and pond culture techniques are sufficiently developed to allow initiation of farming
operations throughout many areas of the southeastern U.S. Although aquaculture of these fish
appears economically attractive, there are certain impediments to large-scale culture. Such im-
pediments include issues of seed stock availability and cost, and restrictive laws and regulations.
Such issues are being addressed and progress has been achieved in alleviating some of the im-
pacts of these factors. It is expected that within the near future a sizable industry will be developed
and that cultured striped bass and its hybrids will become readily available in many seafood markets
and restaurants.

Introduction have been highly encouraging and now there is commer-
cial interest in many states 'to rear these fish for market
Native stocks of striped bass, Morone saxattlis, have sup- (Smith and Jenkins 1985a).
ported major fisheries in the United States along the Atlan- This manuscript reviews and discusses the various con-
tic coast. However, since 1973 landings have declined siderations associated with aquaculture of striped bass and
dramatically (Fig. 1) and in an effort to protect the remain- its hybrids in North America. In particular, information
ing stocks, commercial fishing bans have been implemented on broodstock acquisition and culture, spawning, hatching,
in many states. The striped bass is also an important larval rearing, and production data are presented. Also,
recreational species and has been stocked in lakes, reser- comparative performance data on some of the various
voirs, and rivers throughout the United States to support hybrid striped bass crosses are include&
sport fisheries (Stevens 1984).
The striped bass is well known. in the marketplace where
it commands a high price (Sport Fishing Institute 1984; Status of Aquaculture
Swartz 1984). Because of the scarcity of this species and Supply of "Seed Stock"
its high market value, interest has increased for culturing
striped bass or a suitable market substitute. During recent A basic impediment to the large-scale development of
years, research and development activities have included striped bass and hybrid bass farms has been the lack of
attempts to rear striped bass and its hybrids in ponds, net- a dependable supply of seed stock (joint Subcommittee on
pens, raceways, and tanks (Powell 1973; Valenti et al. Aquaculture 1983; Smith andjenkins 1985b; Smith 1987).
1976; Wawronowicz and Lewis 1979; Williams et al. 1981; Presently, state and federal hatcheries rely on the capture
Kerby et al. 1983a, b; Woods et al. 1983, 1985; Collins of wild adults from spawning grounds to support hatchery
et al. 1984; Kerby et al. 1987; Smith et al. 1987). Results operations (Harrell 1984). However, the private sector is
typically prohibited. from collecting broodstock using the
same techniques (especially electrofishing) and from ob-
'Contribution Number 233 from the South Carolina Marine Resource taining stock from areas used by these public hatcheries.
Center. Consequently, -acquisition of ripe broodstock by the private
'Preparation of this report was supported in part by Dep. of Commerce,
NOAA, Office of Sea Grant, under Contract Number NA85AA-D- sector is unpredictable, inefficient, and highly regulated.
SG-121 and the State of South Carolina. In recent years, there has also been an overall decrease in

54 NOAA Technical Report NMFS 92

5-
C/)

0
1--- 4--

0
CC 3
F-
LU

1 __

0 . . . . . . . . . . Figure 1
64 66 68 70 72 74 76 78 80 82 84 86 Commercial landings of striped bass on the
YEARS Atlantic coast. (Fisheries of the United States,
formerly Fisheries Statistics of the U.S.)

broodstock numbers such that even state and federal
hatchery managers are currently experiencing difficulty in Table I
meeting their production goals. Age, size, and percent maturity for cultured striped bass
broodstock (updated from Smith and Jenkins 1986).,

Males Females
Broodstock Development Research
Age Mature6 Weight Mature' Weight
In order to alleviate this "seed stock" problem, South (Months) (kg) (kg)
Carolina researchers began in 1982 to examine the feasi-
bility of developing domesticated broodstock to support 21 22 - 0 -
hatchery operations (Smith and Jenkins 1984). Progeny 33 100 2.3 25 3.4
46 100 3.9 75 4.9
of wild striped bass were reared in tanks and subjected to 60 100 5.2 100 6.4
controlled temperature and photoperiod regimes (Smith
and Jenkins 1986). During a five-year period the growth, 'Based on number of males and females at 60 months of age. 19 %
maturity, and spawning success of these fish were mom- of the fish did not mature by age 60 months.
6Fish which expressed milt.
tored. At 33 months of age males had attained a size of 'Fish with eggs greater than 700 jA in diameter.
2.3 kg and essentially all were mature (Table 1). However,
at 3 3 months of age only 25 17o of the females were mature
and difficulty was experienced in spawning these fish
(Smith and Jenkins 1986). During the following two years space required is reduced but effort is intensive and often
additional females matured while no additional mature broodstock die as a result of handling and stripping pro-
males were observed (Table 1). cedures. In South Carolina, both techniques have been
Spawning of wild striped bass broodstock is accom- used with 3- to 5-year old cultured broodstock. It is recom-
plished by injecting the newly captured fish with human mended that the "tank spawning" method be employed
chorionic gonadotropin (HCG) at a rate of 330 I.U./kg. with striped bass in order to reduce fish injury. Although
Then, the fish are either "naturally" tank spawned by pair- substantial progress has been achieved, the culture tech-
ing up males and females in a tank (Bishop 1975), or the niques developed for striped bass females need additional
eggs are stripped from the female at time of ovulation and research to improve the predictability of spawning success.
artificially fertilized with stripped milt (Bayless 1972; Bonn In contrast, development of domesticated male broodstock
et al. 1976). In the first case, larger facilities are needed has been highly successful. Males can be produced in 2
to house spawning tanks but effort is less intensive and to 3 years and used over several years (Smith and Jenkins
damage to broodstock is reduced. In the latter method, the 1986).

Smith: Aquaculture of Striped Bass 55

The techniques described above are used for the pro- Table 2
duction of striped bass fry. However, the striped bass is Summary of reciprocal cross fry production using captive
not the preferred fish for an aquaculture operation because wild white bass females and cultured striped bass males (up-
of its slower growth during the first two years and its dated from Smith and Jenkins 1986).
reduced environmental tolerances. Concurrent research
conducted with the striped bass and white bass (M. @hrysops) Fish Weight Larvae Hatched
crosses has indicated that these hybrids are the preferred Trial No. (kg) (no.)
fishes for aquaculture development (discussed later in text) ' 1 0.63 33992
In order to make "original cross" hybrids, female striped 2 0.46 37247
bass and male white bass must be stripped because they 3 0.64 84185
can not be induced to tank spawn together. Consequently, 4 0.61 17870
the problems of acquisition, predictability, and reuse of 5 0.64 87950
6 0.68 48 100'
female striped bass occur. However, the "reciprocal cross" 7 0.73 48 100'
(female white bass x male striped bass) can also be per- 8 0.65 48 100a
formed which results in a hybrid which is quite satisfac- 9 0.66 481001
tory for aquaculture development (see later section). White 'All hatches were pooled, 192000 fry produced.
bass are smaller than striped bass at maturity (minimum
size for females -300 g vs. -3.4 kg for striped bass) and
this species is abundant throughout many areas of the U.S.
Although this species is of recreational importance, there can be left in the spawning tank to hatch or the eggs can
is much less public sentiment associated with their removal be collected and placed in McDonald jars. Striped bass eggs
for private hatchery use compared to the collection of are sernibuoyant and are easily kept rolling in the jars. In
striped bass (especially females). contrast, white bass eggs are highly adhesive and will
Broodstock development activities with white bass have readily form large clumped masses of eggs. Such clump-
been based on the use of wild-caught fish, although cultured ing is undesirable as dead eggs are not easily removed and
white bass may also prove to be suitable. Researchers in fungal infections can readily occur. In order to eliminate
South Carolina have demonstrated that adult wild fish can the adhesiveness, the fertilized eggs are placed in the
be captured in the fall, matured in outdoor tank culture McDonald jars and aerated in a tannic acid solution (150
systems using ambient conditions, and spawned in the mg/L) for about 10 minutes before clean fresh water is
spring (Smith andjenkins 1986). Unlike striped bass, adult flowed through the jars. This technique reduces clumping
white bass are either ripe males or ripe females in the spring and results in a higher hatch rate of reciprocal cross hybrids
with few or no fish of unknown sex. Work-to-date indicates (Charles C. Starling, Florida Game and Fresh Water Fish
that white bass females and striped bass males are easy to Commission, Webster, FL 33597, pers. commun., March
spawn and will naturally mature together in outdoor tanks 1986).
(Smith and Jenkins 1985b, 1986, 1987). HCG is also used
with white bass but it is administered at a higher level Fingerling Production
(7 50-1500 1. U. /kg). Average production of reciprocal cross
hybrid bass fry is in the range of 35 000 to 90 000 per female The fry are reared in either freshwater or brackish water
fish 460 to 640 g in weight (Smith and Jenkins 1986; Table nursery ponds (Phase I) to a small juvenile size over a
2). Recent demonstration of these spawning techniques 30-45 day period (Bonn et al. 1976; Geiger 1983a, b;
using captive, wild broodstock should enhance the develop- Parker and Geiger 1984). During the last 2-3 weeks of
ment of hybrid striped bass aquaculture farms (Smith and nursery rearing, dry feed is added to the ponds to serve
Jenkins 1987). as a replacement for the diminishing supply of zooplank-
ton of a suitable size. Typical stocking density for the
Incubation and Hatching Phase I nursery ponds is about 250000 fry/ha. Survival
ranges from 0 to about 80%, but 25 Olo is common (Parker
Eggs of striped bass and its hybrids are typically incubated and Geiger 1984). Temperature in the ponds ranges from
in McDonald hatchingjars (Bonn et al. 1976). Depending about 18 to 25'C. During 1985 and 1986, nursery trials
on egg type, approximately 100000-250000 eggs are were conducted at the S.C. Wildlife and Marine Resources
placed in each jar. Water injected through a center tube Department's Waddell Mariculture Center (WMC). Pro-
causes continuous upwelling and oxygenation, and eggs duction from these trials conducted at dissimilar densities
hatch in about 48 hours at temperatures of 18' to 20'C, was 281000/ha for striped bass and 110300/ha on the
(Bayless 1972). At time of hatching, fry swim up with the average for hybrids (Table 3).
outflowing water and are concentrated in outside aquaria. The small juveniles are harvested by draining the pond
In situations where striped bass are tank-spawned, eggs and collecting the fish in catch basins adjacent to the drain

56 NOAA Technical Report NMFS 92

Table 3
Results of Phase I nursery trails with striped bass and its hybrids in brackish water (salinity
4-8 ppt) ponds.

Stocking data Harvest data

Density Duration Density Survival Mean wt.
Type of fish (no./ha) Date (d) (no./ha) M (g)

Striped bass 600000 4/28/87 30 281630 46.9 0.36
F, original hybrid 128000 5/08/85 41 96000 75.0 0.45
F, reciprocal hybrid 300000 3/10/86 57 74500 24.8 2.20
F, reciprocal hybrid 300000 3/10/86 56 161030 53.7 0.67

structure. Fish produced by private hatcheries are then eggs (Stevens 1966, 1967). The focus of the hybridiza-
graded and restocked in Phase 11 rearing ponds at a lower tion work was to develop a fish that had the growth and
density or sold to other farmers. Indoor intensive nursery feeding characteristics of the striped bass and the envi-
techniques using tanks have also been developed for ronmental adaptability and less stringent spawning re-
Phase I rearing of striped bass and its hybrids (Lewis and quirements of white bass (Bonn et al. 1976). The first
Heidinger 1981; Kerby et al. 1983a; Carlberg et al. 1984; hybrid produced was the "original cross." Later other
Smith and Jenkins 1984). However, production ofjuveniles crosses were made using white perch (M. americana) and
from such tank nursery systems is often highly variable and yellow bass (M. mississippiensis) males as well as reciprocal
the pond-rearing method is currently the preferred tech- and backcrosses (Bayless 1972). Early evaluation of these
nique for Phase I production. various hybrids was somewhat opportunistic and not well
controlled. Nevertheless, from these studies it was con-
Performance Characteristics cluded that the original cross hybrid was a good fish and
of Striped Bass and Hybrids stocking programs were initiated (Bishop 1968; Logan
1968; Williams 1971; Ware 1975). Field observations in-
Hybridization studies with striped bass were initiated in dicated that the hybrids outgrew striped bass during the
1965 after it was demonstrated that HCG could be used first two years and were both easier to produce and har-
to induce final maturation and ovulation of striped bass dier than striped bass.

Table 4
Summary of indoor tank studies examining performance characteristics of striped bass and its hybrids (updated from Smith
and Jenkins 1985b).

Fish stocked Harvest data

Size Density Duration Size Survival Biomass Feed
Type (g) (no. /m') (d) (g) M (kg/m') conversion

Study number 1
Striped bass 4.4 31.6 287 289 96 8.8 2.3
F, original hybrid 6.4 31.6 287 507 100 16.0 2.2
F2 hybrid bass 12.8 31.6 287 347 84 9.2 2.7
Study number 2
F, original hybrid 23.0 31.6 140 263 99 8.3 1.9
F, white perch hybrid 21.0 31.6 140 176 99 5.5 2.5
Study number 3'
Striped bass 6.0 29.2 56 34 89 1.4 1.7
F, original hybrid 13.2 29.2 56 77 98 3.6 1.6
F, reciprocal hybrid 10.9 29.2 56 57 93 2.5 1.9
Backcross hybrid 21.1 29.2 56 71 92 3.1 2.4

'Study is currently in progress.

Smith: Aquaculture of Striped Bass 57

200

1-80

160 Reciprocal Hybrids

140--

120

100 original Hybrids
CD
LU
so--

60-

40-

20- Figure 2
Comparison of F, origi-
0 t nal and reciprocal cross
0 28 56 84 1 12 126 140 168 184 hybrids reared in ponds at
TIME (days) a density of 10000 fish/
ha.

With the increasing interest in aquaculture, more (T. Smith, unpublished).
detailed performance information was desired for the In summary, the hybrids of striped bass and white bass
various hybrids. Controlled studies were undertaken appear well suited for aquaculture. They can be raised in
primarily in South Carolina and North Carolina. To date, fresh or brackish water (Smith et al. 1986) and exhibit high
not all striped bass hybrids have been tested but sufficient growth and survival rates. Further, they can be reared in
information is available to identify suitable candidates for a broad range of water temperatures and are not killed by
aquaculture use. low (7'C) or high (33'C) water temperatures (Smith et al.
A number of comparative studies have been undertaken 1987).
by the Charleston Laboratory, S.C. Wildlife and Marine
Resources Department, using indoor tanks which can recir- Grow-Out Production
culate either fresh or brackish water. In the first study
striped bass were compared to reciprocal cross hybrids and Research data from production trials to produce market-
to F2 hybrids (F1 original cross x F, original cross)@ size fish (>,568 g) have been limited although there is a
Results of this 287-day study indicated that the reciprocal commercial operation producing hybrids in tanks using
cross hybrid grew fastest and had a high survival rate geothermal water in California and another company is
(Table 4, Smith et al. 1985). In the second study the using net pens to produce striped bass in New York. Un-
original cross hybrid was compared to a striped bass x fortunately, production data from these two operations are
white perch hybrid. Again, results indicated that the white proprietary. In 1984, North Carolina researchers harvested
bass hybrid grew rapidly and had a high survival rate grow-out trials in 0. 1 ha earthen ponds using original cross
(Table 4). In a current study, the original and reciprocal hybrids. Survival averaged 83.9% and production ranged
cross hybrids are outperforming the striped bass. However, from 5 247 to 5 765 kg/ha (mean 5504 kg/ha) (Kerby et al.
a backcross hybrid (F1 original x striped bass male) is 1987). Recently, the S.C. Wildlife and Marine Resources
also performing well at the present time and may be an completed its first pond grow-out trial using original cross
additional aquaculture candidate (Table 4) (T. Smith, un- hybrids. This study was conducted in a 0.5 ha pond at
published). Results of these replicated tank studies are also WMC. One-year-old juveniles (mean size 220 g) were
being corroborated in pond culture trials (Figs. 2, 3) stocked in March 1986 at a density of 12000/ha for final

58 NOAA Technical Report NMFS 92

90--

80--

70-- Reciprocal Hybrids

60-

50-
F_

0
Lu

30-

20-
Striped Bass

Figure 3
0 1 Hill 1 HHHH! .... ... Comparison of striped
bass and reciprocal cross
12 24 36 48 60 72 84 96 hybrids reared in ponds at
TIME (days) a density of 37000 fish/
ha.

grow-out to market size. During the growing season Market Testing
(April-November), fish were fed a commercial trout pellet
twice daily and satisfactory water quality conditions were The opportunities and constraints associated. with the
maintained with paddlewheel aerators and water exchange. marketing of hybrid striped bass were recently examined
From December, 1986 to January, 1987 the fish were har- by Carlberg and Van Olst (1987). Prices received for the
vested and marketed. At harvest, mean fish size was 755 g cultured hybrids have been in the range of $5.50-1 1.00/kg
and survival was 93 To (Table 5). Total production was depending on product type (e.g., iced, gutted, live) and
8 323 kg/ha with 93 7o of the fish > 568 g (1 Y4 lb). These specific market (e.g., wholesaler, retailer, restaurant).
results are highly encouraging and represent the highest Organoleptic testing of pond-reared cultured hybrids has
pond production level yet achieved. Currently, a pond been conducted by the Southeast Fisheries Center
grow-out trial is underway with reciprocal cross hybrids [National Marine Fisheries Service (NMFS)], Charleston,
and growth to the present appears similar. From an aqua- S.C. Results indicate that the hybrid striped bass is a mild
culture perspective, routine production levels of 3900 to flavored fish which should have good market appeal
7 800 kg/ha should be attainable using techniques similar (Michael jahncke, NMFS, Charleston, SC 29412, pers.
to those employed for producing channel catfish (Malurus commun., June 1987). Restaurant testing has also been
nebulosus). conducted using our pond-reared hybrid bass. Again,

Table 5
Stocking and harvest data for pond grow-out trial using original cross hybrid striped bass.

Stocking data Harvest data

Density Mean wt. Age of fish Pond size Duration Mean wt. Survival Production Feed
(no./ha) (9) (d) (ha) (d) (g) (%) (kg/ha) conversion

2000 220.2 327 0.6 240 755 93 8323 1.6

Smith: Aquaculture of Striped Bass 59

Table 6
Consumer evaluation (N = 216) of hybrid striped bass served in two Flordia restaurants (adapted from Liao et al. 1987).
Rating scale: 1 = poor, 2 = fair, 3 = good, 4 = very good, 4 = excellent.

Evaluation catergories Consumer repurchase (01o)

Restaurant Entree Price Appearance Texture Smell Taste Price No Maybe Yes

A Broiled 10.95 4.7 4.6 4.8 4.7 4.4 2 4 94
Fried 10.95 4.2 4.7 4.2 4.2 4.1 22 0 78
Grilled 10.95 41. 4 4.4 4.4 4.3 4.3 8 0 92

B Broiled 4.95 4.3 4.2 4.4 C 5 4.2 2 5 93
Fried 4.95 4.1 4.2 4.1 4.4 4.2 2 5 93

results were favorable indicating that these fish are highly Conclusions
acceptable to consumers (Table 6, Liao et al. 1987). Thus, Commercial farming of striped bass hybrids is an emerg-
cultured hybrid striped bass appear to be an excellent ing industry in the United States which will need additional
market substitute for wild striped bass. research and extension activities to reach its full potential.
Pond culture technology is currently available and is now
Aquaculture Constraints in the process of being transferred to the private sector.
At the present time, there are two major constraints to Indeed, during the past year cooperative demonstration
the development of striped bass farms: 1) seed stock projects have been initiated in several states including
availability and cost, and 2) laws and regulations. Recent Maryland and North Carolina. Within the next 3-5 years,
research in South Carolina on the domestication and pond culture operations are expected to develop throughout
culture of broodstock has demonstrated commercially the southeast and middle Atlantic states as well as in
practical techniques for controlled production of "seed California.
stock. " Thus, in the near future, private hatcheries Although the research data base appears adequate for
should be able to develop their own captive broodstock development of the industry, additional study is needed
(especially striped bass males and white bass females) in the areas of 1) broodstock development and genetics,
and thereby increase the predictability and availability of 2) identification of other suitable hybrids, 3) nutrition,
fry and small juveniles. This use of captive broodstock may 4) market development, 5) disease treatment and preven-
also result in lower production costs and lower seed stock tion, and 6) laws and regulations. All these areas can
costs. substantially impact the economics of bass aquaculture. For
The legal issues have been a serious constraint in example, profitability could be improved by development
of hybrids exhibiting faster growth through genetic breed-
many states where fishery bans were instituted to protect ing or manipulation (induced polyploidy) and by the iden-
native fishery stocks including game species such as striped tification of more cost-effective rations.
bass. In most cases, there was no exclusion for farm-reared The future appears highly promising for the develop-
fish. During the past year there has been substantial ment of hybrid striped bass farms throughout many areas
interest expressed by the agricultural community, busi- of the United States. For the most part, these fish will be
nessmen, and landowners, to commercially farm striped reared as a high quality seafood product. However, it is
bass and its hybrids. Consequently, pressure has been also expected that these fish will be used to support recrea-
exerted on legislators to legalize the culture and sale of these tional fee-imposed fishing operations as well. Such opera-
fish with the result that many states are currently re- tions will provide income to the operator while at the same
examining their laws and making provisions for bass time affording recreational fishing opportunity to the public
aquaculture (e.g., Florida, Virginia, North Carolina, sector. In summary, the striped bass x white bass hybrids
Georgia, Mississippi). In addition to the laws concerning are not only important recreational fish but they also
possession and sale of fish, there are many additional appear to be excellent candidates for commercial
regulations which affect the aquaculturist. These deal with development.
site permits, discharge constraints, use of medications, and
broodstock collection techniques, among others Uenkins
1987). Some of these laws and regulations are also being Citations
re-examined because of their impact on aquaculture Bayless, J.D.
development. 1972. Artificial propagation and hybridization of striped bass,

60 NOAA Technical Report NMFS 92

Morone saxalilis (Walbaum). South Carolina Wildl. Mar. Re- bass. Presented at the Symposium on Markets for Seafood and
sources Dep., 135 p. Aquacultural Products, Charleston, S.C., August 1987. (Mimeo
Bishop, R.D. report.) S.C. Wildl. and Mar. Resources Dep., Charleston,
1968. Evaluation of the striped bass (Roccus saxatilis) and white bass SC.
(Roccus chrysops) hybrids after two years. Proc. Annu. Conf. Logan, H.J.
Southeast. Assoc. Game Fish Comm. 21:245-254. 1968. Comparison of growth and survival rates of striped bass x
1975. The use of circular tanks for spawning striped bass (Morone white bass hybrids under controlled environments. Proc. Annu.
saxatilis). Proc. Annu. Conf. Southeast. Assoc. Game Fish Conf. Southeast. Assoc. Game Fish Comm. 21:260-263.
Comm. 28:35-44. Parker, N.C., andJ.C. Geiger.
Bonn, E.W., W.M. Bailey, J.D. Bayless, K.E. Erickson, and 1984. Production methods for striped bass. In Third report to the
R.E. Stevens. fish farmers (H. K. Dupree and J. V. Huner, eds.), p. 106-118.
1976. Guidelines for striped bass culture. Striped Bass Commit- U.S. Fish Wild]. Serv., Washington, D.C.
tee of the Southern Division. Am. Fisheries Soc., Bethesda, MD, Powell, M.R.
103 p. 1973. Cage and raceway culture of striped bass in brackish water
Carlberg, J.M., and J.C. Van Olst. in Alabama. Proc. Annu. Conf. Southeast. Assoc. Game Fish
1987. Processing and marketing. In Hybrid striped bass culture: Comm. 26:345-356.
status and perspective (Hodson et al., eds.), p. 73-82. Univ. Smith, T.I.J.
North Carolina Sea Grant Publ. No. 87-03. 1987. Hatchery. In Hybrid striped bass culture: status and
Carlberg, J.M., J.C. Van Olst, MJ. Massingill, and T.A. Hovanec. perspective (Hodson et al., eds.), p. 17-22. Univ. North Carolina
1984. Intensive culture of striped bass: a review of recent tech- Sea Grant Publ. No. 87-03.
nological developments. In The aquaculture of striped bass - a Smith, T.I.J., and W.E. Jenkins.
proceedings (J.P. McCranen, ed.), p. 89-127. Univ. Maryland 1984. Controlled spawning of F, hybrid striped bass (Morone sax-
Sea Grant Publ. No. UM-SG-MAP-84-01. alilis x M. chrysops) and rearing of F2 progeny. J. World Mari-
Collins, C.M., G.L. Burton, and R.L. Schweinfor-th. culture Soc. 15:147-161.
1984r. High density culture of white bass x striped bass fingerlings 1985a. Status of aquaculture of striped bass (Morone saxatilis) and
in raceways using power plant heated effluents. In The aqua- its white bass (M. chrysops) hybrids and research in South Carolina.
culture of striped bass - a proceedings (J.P. McCraren, ed.), p. In Proceedings of the 2nd int. conf. warm water aquaculture - fin-
129-142. Univ. Maryland Sea Grant Publication No. UM-SG- fish. Brigham Young Univ. p. 553-582.
MAP-84-01, 262 p. 1985b. Aquaculture research with striped bass (Morone saxatilis) and
Geiger, J.G. its hybrids in South Carolina. Proc. Annu. Conf. Southeast.
1983a. Zooplankton production and manipulation in striped bass Assoc. Fish Wild]. Agencies 39:219-227.
rearing ponds. Aquaculture 35:331-351. 1986. Culture and controlled spawning of striped bass, Moronesax-
1983b. A review of pond zooplankton production and fertilization atilis, to produce striped bass, and striped bass x white bass, M.
for the culture of larval and fingerling striped bass. Aquaculture chrysops hybrids. Proc. Annu. Conf. Southeast. Assoc. Fish WiIdl.
35:353-369. Agencies 40:152-162.
Harrell, R.M. 1987. Controlled spawning of cultured striped bass, Moronesaxatilis.
1984. Review of striped bass broodstock acquisition, spawning (Abstract.) 117th Annual Meeting Am. Fisheries Soc.
methods and fry production. In The Aquaculture of striped bass Smith, T. I.J., W. E. Jenkins, and J. F. Snevel.
- a proceedings (J.P. McCraren, ed.), p. 45-57. Univ. Maryland 1985. Production characteristics of striped bass, and F, and F2
Sea Grant Publ. No. UM-SG-MAP-84-01. striped bass/white bass hybrids reared in intensive tank systems.
Jenkins, W.E. J. World Mariculture Soc. 16:57-70.
1987. Laws and regulations. In Hybrid striped bass culture: status Smith, T.I.J., W.E. Jenkins, and R. Haggerty.
and perspective (Hodson et al., eds.), p. 93-100. Univ. North 1986. Growth and survival ofjuvenile striped bass (Moronesaxatilis)
Carolina Sea Grant Publ. No. 87-03. x white bass (M. chrysops) hybrids reared at different salinities.
joint Subcommittee on Aquaculture. Proc. Annu. Conf. Southeast. Assoc. Fish Wild[. Agencies 40:
1983. Striped bass species plan. In National aquaculture develop- 1+3-151.
ment plan, Vol. 2, p. 136-11,5. The joint Subcommittee on Smith, T.I.J., W.E. Jenkins, A.D. Stokes, and R.A. Smiley.
Aquaculture of the Federal Coordinating Council on Science, 1987. Pond production trials with striped bass (Morone saxatilis) and
Engineering, and Technology, Washington, DC. white bass (M. chrysops) hybrids. (Abstract.) J. World Aquaculture
Kerby, J.H., L.C. Woods 111, and M.T. Huish. Soc. 18(l):8A.
1983a. Culture of striped bass and its hybrids: a review of methods, Sport Fishing Institute.
advances and problems. In Proceedings of warm water fish 1984. An uncertain future. Sport Fishing Inst. Bull. 35(l):1-5.
culture workshop (R.R. Stickney and F.P. Myers, eds.), p. 23-54. Stevens, R.E.
World Maricult. Soc., Spec. Publ. No. 3. 1966. Hormone-induced spawning of striped bass for reservoir
1983b. Pond culture of striped bass x white bass hybrids. J. stocking. Prog. Fish-Cult. 28:19-28.
World Maricult. Soc. 14:613-623. 1967. A final report on the use of hormones to ovulate striped bass
Kerby, J.H., J.M. Hinshaw, and M.T. Huish. Roccus saxatilis (Walbaum). Proc. Annu. Conf. Southeast. Assoc.
1987. Increased growth and production of striped bass x white Game Fish Comm. 18:525-538.
bass hybrids in carther ponds. J. World Aquacult. Soc. 1984. Historical overview of striped bass culture and manage-
18(l):35-43. ment. In The Aquaculture of striped bass: a proceedings (J.P.
Lewis, W.M., and R.C. Heidinger. McCraren, ed.), p. 1-5. Univ. Maryland Sea Grant Publ. No.
1981. Tank culture of striped bass: production manual. Illinois UM-SG-MAP-84-01.
Striped Bass Project IDC F-2, 6-R, Univ. Carbondale, Illinois, Swartz, D.
115 p. 1984. Marketing striped bass. In The Aquaculture of striped bass:
Liao, D.S., T.I.J. Smith, and W.E. Jenkins. a proceedings (J.P. McCraren, ed.), p. 233-254. Univ.Mary-
1987. Preliminary market analysis for cultured hybrid striped land Sea Grant Publ. No. UM-SG-MAP-84-01.

Smith: Aquaculture of Striped Bass 61

U.S. Department of Commerce. Fish-Cult. 41:138-140.
1978-87. Fisheries of the United States. (Formerly Fishery Statis- Williams, H.M.
ticsoftheU.S.) U.S. Dep. ofCommer., NOAA,NMFS,var. 197 1. Preliminary studies of certain aspects of the life history of
pagination. the hybrid (striped bass x white bass) in two South Carolina reser-
U.S. Department of Interior. voirs. Proc. Annu. Conf. Southeast. Assoc. Game Fish Comm.
1965-77. Fishery Statistics of the U.S. U.S. Dep. Interior, Fish 24:424-431.
Wild]. Serv., var pagination. Williams, J.E., P.A. Sandifer, and J.M. Lindberg.
Valenti, R J., J. Aldred, and J. Liebell. 198 1. Net-pen culture of striped x white bass hybrids in estuarine
1976. Experimental marine cage culture of striped bass in northern waters of South Carolina: a pilot study. J. World Maricult. Soc.
waters. Proc. World Maricult. Soc. 7:99-108. 12(2):98-110.
Ware, Fj. Woods, L.C. III, J.H. Kerby, and M.T. Huish.
1975. Progress with Morone hybrids in fresh water. Proc. Annu. 1983. Estuarine cage culture of hybrid striped bass. J. World
Conf. Southeast. Assoc. Game Fish Comm. 28:48-54. Maricult. Soc. 14:595-612.
Wawronowicz, L.J., and W.M. Lewis 1985. Culture of hybrid striped bass to marketable size in circular
1979. Evaluation of the striped bass as a pond food fish. Prog. tanks. Prog. Fish-Cult. 47(3):147-153.

Computerized Image Analysis for Selective Breeding of Shrimp:
A Progress Report

L. JAMES LESTER, KIM S. LAWSON, and MARK J. PIOTROWSKI

Aquaculture Genetics Laboratory
University of Houston - Clear Lake
2700 Bay Area Blvd.
Houston, TX 77058

TENG-CHEONG B. WONG

Research Institutefor Computing and Information Systems
University of Houston - Clear Lake
2700 Bay Area Blvd.
Houston, TX 77058

ABSTRACT

Genetic improvement of aquaculture organisms will be an important component of future
progress by the aquaculture industry. The benefits of selective breeding could be obtained more
rapidly if high selection intensity could be applied. This will be practical with an approach based
on computerized image analysis which can provide an accurate estimate of organism size.
Preliminary studies were conducted with a computer assisted morphometric analysis which ob-
tains more information on size and shape from an image of a shrimp body than more conven-
tional direct measurement. Using computer digitization, a truss network of dimensional variables
was collected from photographs of three species of adult penacid shrimp and used in canonical
discriminant analysis.This approach was found to provide more information for discrimination
than a typically labor intensive conventional morphometric method. The results demonstrate the
usefulness of two-dimensional images for selection or classification. A truss network data set was
also collected from full-sib families of juvenile shrimp using the new method. This study shows
the application of the technique to images from living animals in a format resembling selection.
Phenotypic variation in size is analyzed by analysis of variance, and heritabilities of size based
on principal component scores and single variables are calculated. The multivariate estimator
of size had more of the variance in size attributed to family differences than most single vari-
ables (h' = 0.60 for PC scores vs. a mean of 0.49 for single variables). Ways in which com-
puterized image analysis might be employed in selective breeding of shrimp are discussed.

Introduction intensive characteristics of such a program will benefit
greatly from the use of computer image analysis, a valuable
There has been much discussion about the possible use of alternative to human observers. Third, obtaining accuracy
high selection intensities for aquaculture animals. High in the selection process will require the use of multivariate
selection intensities are possible because the very high procedures in the assessment of organism size.
fecundities of aquacultured organisms (e.g., penaeid If we compare the fecundity of shrimp with even the most
shrimp) permit a very small fraction of the individuals to fertile domestic animal, they are different by orders of
be used for replacement of broodstock. This paper will sup- magnitude. In commercial shrimp culture facilities, female
port several premises related to the use of high selection Penaeus vannamei have been observed to average 60000
intensities in a broodstock selection program. First, the im- nauplii per spawn and 8-10 spawns during their produc-
plementation of such selection intensities will be possible tive life in a maturation facility (B. Ribelin, Laguna Madre
and should not result in genetic drift. Second, the labor Shrimp Farm, P.O. Box 4043, Los Fresnos, TX 78566,

64 NOAA Technical Report NMFS 92

pers. commun., June 1986). Let us use the figure of 600 000 tical (i.e., high selection intensities would require major
larvae from an average female and follow the offspring investments of personnel and time).
through a production cycle, assuming 50% survival in each In a previous publication (Lester 1983), it was recom-
stage. This percentage is low, but not outside the range mended that a single abdominal measurement, sixth seg-
of observed values. The numbers would change as follows: ment depth (SSD), be used as a criterion in selection for
300 000 postlarvae, 150 000 juveniles, and 75 000 harvested size of penaeid shrimp (Fig. 1). Several preliminary selec-
subadults. Assuming a survival of 50% from selection to tions have been performed using this character. A single
reproduction, one would only need to select four individual human observer is capable of applying this criterion to
subadults to replace the original two breeders. 300-500 shrimp per hour, but not for many hours con-
Selection intensity (0 is related to the response of a tinuously (pers. observ.). Nevertheless, at a selection in-
population to selection (R) by the simple equation R = tensity of 3.0, one would be required to measure approx-
I.op h2 in which h2 is the heritability of the trait under imately 300 shrimp, about 40-60 minutes of work, to
selection and op is the phenotypic variance of the trait. select a single potential broodstock. In order to obtain a
The relationship between 1'and the proportion of animals broodstock of 3000, which is not large by commercial stan-
selected (p) for truncation selection of a normally distrib- dards, the observers would have to work for over 2000
uted trait is i = z1p, where z is the height of the normal hours. A microcomputer imaging system could perform
curve at the truncation point (Falconer 1981). Thus the the measurements and make a decision on whether to cull
maximum selection intensity for the above example (based or select in several seconds. Computer analysis would be
on a proportion of 4 selected from 75000) would be ap- limited by how fast the shrimp could be moved through
proximately 4.3, if all the survivors could be measured. the imaging system. A single unit could measure con-
Lasley (1978) gives the percentage of the progeny re- tinuously and accurately for any amount of time necessary.
quired for broodstock replacement in domestic animals as No matter what criterion is employed for the selection
4-5 % of males and 40-50 % of females in beef cattle; 1-2 % process, there will be an associated measurement error. At
of males and 10-15% of females in swine and chickens. a high selection intensity, the variance due to measurement
Such broodstock selection programs for domestic animals error could exceed the true phenotypic variance of the
obtain selection intensities of 2.0 to 2.7 for males and 0.6 potential broodstock in the distribution. This would repre-
to 0.8 for females (Falconer 1981). With the very small pro- sent a serious problem for the classification of animals into
portion of the population that is required for replacement potential broodstock and culls. This factor in misclassifi-
broodstock in aquaculture species, it is possible to achieve cation can be reduced by moving from univariate to
selection intensities of 3.0-4.0. Current selection programs multivariate classification. Thus we are recommending that
at commercial shrimp farms select the top 1/2 to 1/3 of the high intensity selection not be based on a single measure-
size distribution. If we assume that the trait under selec- ment such as SSD, but rather on a set of measurements.
tion is heritable, the predicted response under high selec- The characteristic of commercial importance, rate of
tion intensity is 3 to 4 times higher than under the current gain in size, is a latent variable which can not be measured
system. This could be accomplished using within-family directly in these selection programs. It can be estimated
selection to remove the danger of bottlenecks caused by by analysis of multiple variables which are correlated to
unrecognized restriction of the gene pool and inbreeding. it. (See Bookstein et al. 1985, for a thorough discussion
The limitations on selection intensity are primarily logis- of multivariate estimation of the latent variable size.) In

AAC
PCL Figure I
FLF SAD FSL CW Diagram of shrimp with measurements used in the first
L study. Description of measurements: PCL = posterior
> margin of the orbit to posterior edge of carapace;
CW = carapace width at the level of the last dorsal
rostral tooth; FSL = first abdominal segment length;
AAC = circumference at the intersection of the second
and third abdominal segments; SAD = depth at the in-
tersection of the second and third abdominal segments;
PAC FLF = fifth segment length with the abdomen maximal-
S ly flexed; PAC = circumference at the intersection of
the fifth and sixth segments; SSL = sixth segment
length; SSD = depth at the midpoint of the sixth
segment.

Lester et al.: Image Analysis for Breeding Shrimp 65

DM4 DM3 DM2 DMI
Figure 2
Diagram of shrimp with landmarks used in the adult
DM5 truss network study. Descriptions of points: DN11 =
Posterior rostral tooth; DM2 = dorsal, posterior point
DM6 on carapace; DM3 = dorsal, posterior point on segment
one; DM4 = dorsal, posterior point on segment three;
DN15 = dorsal, anterior point on segment six; DN16
= dorsal, posterior point on segment six; VM1 =
center of antennal basal segment at insertion; VM2
VM1 - center of basal segment of fifth periopod at insertion;
VM3 = center of basal segment of first pleopod at in-
VM6 VM2 sertion; VM4 = center of basal segment of third pleopod
V 4 at insertion; VM5 = center of basal segment of fifth
VM5 VM3 pleopod at insertion; VM6 = ventral, posterior point
on segment six.

our approach to estimation of growth rate, the latent variables which would permit the computer to decide
variable size, is represented by the first principal compo- whether the animal should be selected or culled. At the
nent (PC I) score obtained from a set of size-correlated present time, we are still at an early stage in the develop-
dimensions taken from each shrimp. As variables are ment, using human observers to digitize still images.
added, the variance-covariance matrix gets larger, but the This report deals with initial analyses of two data sets.
variance of the multivariate mean (i.e., mean size) These analyses are based on still images, not video, and
decreases according to the factor 1 - R2 (R2 = the surn only one uses living animals for the images. The first data
of the squared partial correlations between the variables). set was obtained from adults carefully prepared for photo-
In other words, the multivariate data set becomes a better graphing to test the truss network approach on discrimina-
predictor of the underlying variable size when multiple tion of Penaeus species. The second data set was collected
variables correlated with size are used in the estimation from an experiment on growth differences among families
process. of juvenile shrimp. It will be used to examine the way in
There is a school of morphometric studies centered which the partitioning of the variance in size is affected
around the work of F.L. Bookstein (Strauss and Bookstein by changing from a single to a multivariate approach.
1982; Bookstein et al. 1985) which has developed an ap-
proach to the multivariate description of size and shape
for fish. The approach is based on obtaining a set of Materials and Methods
variables that contains much information about specific Testing the Truss Network Approach
regions of the body and is homologous across conspecifics
of different sizes and across closely related species. This Samples of adults were obtained from the following sources
is accomplished by selecting landmark points on the mor- for both data sets: Penaeus seliferus from the wild near Gal-
phology and arranging them into a set of boxes connected veston, TX; Penaeus vannamei collected from harvests of
by all possible nearest neighbor connections into a truss ponds at the King Ranch near Corpus Christi, TX, at the
network as shown in Figure 2 (Strauss and Bookstein 1982; Laguna Madre Shrimp Farm near Harlingen, TX, and
Bookstein et al. 1985). The resultant network consists of at Agromarina de Panama (only in the truss network data
a set of dimensional variables that covary with the latent set) near Aguadulce, Panama; Penaeus stylirostris collected
variable size and contain information about shape-related from harvests of ponds at the King Ranch near Corpus
deformational changes among a set of individuals. At the Christi, TX, and tanks at Marine Culture Enterprises near
present time, we are interested in using this approach to Laie, HI. The standard morphometric study used 104
precisely estimate genetic differences in size. Later we hope P. setiferus, 92 P. stylirostris, and 134 P. vannamel'. The truss
to apply it to the inheritance of shape. network study used 49 P. setiferus, 42 P. stylirostris and 83
One objective of the research program that we have P. vannamel'. All specimens were preserved by freezing and
started was the development of a system that would thawed for photography.
automate broodstock selection. This program would use The standard morphometric data set contained nine
a video camera to obtain an image that would be converted dimensional variables and two weights, head weight and
into digital format. This would be measured and analyzed tail weight. Dimensional values were obtained with dial
by a microcomputer to obtain a summary of multiple calipers except in the case of circumferences which were

66 NOAA Technical Report NMFS 92

obtained by passing a string around the shrimp and The postlarvae were weaned from Artemia to a prepared
measuring the length of string with a ruler. The dimen- diet from the Texas A&M Shrimp Mariculture Project,
sional variables are shown in Figure 1. Data sets of this and reared under the same temperature and salinity con-
type have been obtained from six species of penaeid shrimp. ditions for 40 days on a 14L: I OD light cycle. The juveniles
Only the species P. setiferus, P. stylirostris, and P. vannamei in each cage were fed once per day at ad libitum levels:
will be used here for purposes of comparison to truss net- 50 mg/cage in weeks one and two, 100 mg for five days,
work data sets. 200 mg for five days, 600 mg for five days, and, following
The truss network variables are shown in Figure 2. This sampling for survival and weight at 28 days, 400 mg to
data set contains 26 dimensional variables, no weights, and 8 cages with survival < 50 To and 600 mg to 14 cages with
was obtained from pictures of the adult shrimp. The adults survival > 50 To for the remaining twelve days. Two cages
were positioned on a light table in the laboratory or on light had no survivors at 28 days.
plastic in the field for pictures taken at Agromarina de After 40 days of growth, the shrimp were harvested and
Panama. They were photographed with a scale in the frame survival was determined. Six randomly sampled individ-
using Kodak Tri-X black and white film. The negatives uals from each cage were placed into a viewing cell filled
were mounted as slides and projected onto a digitizing with water and photographed using color slide film. Black
tablet attached to a Tektronix graphics terminal. The and white film can be used and the negative mounted as
operator used the cross-hairs of a mouse to locate 12 land- a slide. Lighting is critical to the location of landmarks on
mark points shown in Figure 2. Points were digitized in small shrimp. Dual fiber optic illuminators positioned to
order from DM6 to DMI to VMl to VM6. The coor- minimize reflection from the exoskeleton were found to give
dinates were stored in a data file, then converted to mea- better results than flash lighting. Slides were projected onto
surements by a Pascal program written by one of the a digitizing tablet. The landmarks used were slightly dif-
authors (Wong, T.-C.). ferent from those employed on the adults. The landmarks
All measurements were log transformed before use to were changed from posterior rostral tooth (DMI in Fig.
remove the exponential effects of growth. Both the stan- 2) to ventral intersection of rostrum and carapace; from
dard and the truss network data files were analyzed by antenna] insertion (VM1 in Fig. 2) to anterior ventral point
multivariate statistical programs, PRINCOMP for prin- on edge of carapace; pleopods marked at anterior edge
cipal component analysis and CANDISC for canonical rather than center point (points VM2 through VM5). The
discriminant function analysis (SAS Institute 1985). The coordinates were converted to measurement variables by
principal component analysis was performed to obtain a the same software and the log-transformed measurement
summary estimate of the variation in the latent variable variables were used in the SAS PRINCOMP program.
size. Discriminant function analysis was performed with Principal component scores were stored and ANOVA was
the two data sets to illustrate the increase in information performed using the SAS GLM program. Then the vari-
relevant to species differences contained in the truss net- ance components were estimated with VARCOMP.
work data set. Heritability was estimated from the variance components
according to Becker (1984).
Multivariate Estimation of Heritability
Nauplii from ten separate spawns were shipped to the Results
University of Houston - Clear Lake Aquaculture Genetics Testing the Truss Network Approach
Laboratory from the Laguna Madre Shrimp Farm. Lar-
vae were acclimated to Marine Mix sea water at labora- The correlation among the variables in the standard data
tory temperature and stocked at 100 larvae/cone in three set is high. Measurement PAC (Fig. 1) has the lowest
one-liter Imhoff cones per family. They were treated under average correlation of 0.68 and SSD the highest, 0.88. This
standardized conditions: 28'C, 30 ppt, constant light, daily indicates that these variables are related in their estima-
changes of water containing Chaetocerosgracilis at 100000 tion of size. The first principal component (PC 1) obtained
cells/mL and Tetraselmis chuil' at 30000 cells/mL. After the for the standard data set represents the latent variable
protozoea-3 stage, Artenzia salina nauplii were fed at 3/mL "size" as indicated by the equal loadings (0.28 to 0.35)
(Lester 1988). At the postlarva-2 stage, populations from of all variables (Bookstein et al. 1985). It explains 8417o of
those families with more than 120 survivors were trans- the variance in the standard data set. The second principal
ferred to 10 liter round cages. Three cages were stocked component contains very little information, 0.05 of the
for each family with 40 postlarvae per cage. These 10 liter variance which primarily represents the posterior ab-
round cages had screened openings near their bottom and dominal circumference. The results from discriminant
continuous water flow from the top. They were distributed analysis can most easily be seen in the plots of canonical
among three large tanks (244 cm x 61 cm x 61 cm) which variate I (CAN1) against canonical variate 2 (CAN2).
provided common water sources. Figure 3 shows the plot resulting from analysis of the stan-

Lester et al.: Image Analysis for Breeding Shrimp 67

Table I
6
Length of the sixth abdominal segment as an indicator of
5 size differences among species and data sets.

v- 4
W Data Set
2 Species Standard (cm) Truss (cm)

P. setiferus 1.64 1.38
> 0 P. stylirostris 1.48 1.73
_j P. vannamei 1.87 1.56

2-2
0.3
Z
-4 S
At this point, there appears to be little on which to choose
the truss network over the standard morphometric vari-
-6 1 ables. However, results of discriminant analysis on the truss
-6 .5 -4 -3 -2 -1 0 1 2 3 4 5 6 network data set are quite different as shown in Figure 4.
CANONICAL VARIATE 2 This analysis results in larger Mahalanobis distances
between the species' centroids, P. setiferus to P. vannarnei
Figure 3 = 5.94, P. seliferus to P. slyhrostris = 6.35 and P. vannamel'
Plot of canonical variates I and 2 from discriminant analysis to P. styllrostris = 5.73. The discrimination of species along
of the standard data set; The species are represented by CANI is primarily due to values of the following variables
ellipses based on the 9517o confidence intervals of the (between group loadings given in parentheses): DM4-DM5
species' centroids. The ellipse for Penaeus setiferus is labeled (O.M), DM4-VM5 (0.92), DM2-DM3 (0.95), DM5-VM5
S, Penaeus stylirostris Y, and Penaeus vannamei V. (0.93), DM3-VM2 (0.97). There are 16 variables with
between-group loadings above 0.75 on CAN2. Five are
measurements from the carapace. Four'are from the sixth
dard data set. The three species show some distinctness, abdominal segment; the remaining seven variables are
but are not well separated. The Mahalanobis distances measurements from the first, second, and third abdomenal
between the centroids of the three species are P. setiferus segments. Thus CAN1 represents the first, fourth and fifth
to P. vannamet = 3.69, P. setiferus to P. stylirostris = 2.51, abdomenal segments and CAN2 the head and the second,
and P. vannarnei to P. styllrostris = 2.32. CANI in Figure third, and sixth abdomenal segments. The canonical cor-
3 shows relatively high loadings of variables AAC (0.86), relations of these discriminant variables to the dummy
FLF (0. 75), SSD (0. 80), CW (0. 73), and PAC (0. 79) which classification variables (species identifiers) are higher than
could be interpreted as a body thickness variable. CAN2 those obtained with the previous data set, for CAN 1, 0. 93
shows high loadings of PCL (0. 88), FSL (0. 80), SSL (0. 78), and for CAN2, 0.91. This indicates that the discriminant
and SAD (0.85). It represents the variation in length of variables derived from the truss network approach are
body segments among species. The goodness of fit of each better for classifying individuals to species than those
discriminant variable to the classification variables (iden- derived from the labor. intensive approach.
tifiers for species) can be evaluated by the canonical cor- It is unfortunate that these data sets are not the result
relation which in the case of CANI is 0.84 and in the case of measurements on the same individuals. However, the
of CAN2 = 0.56. sizes of animals in both data sets are comparable as in-
For the truss network data set, the correlation matrix dicated in Table I in which a variable contained in both
indicated that the correlations were generally lower, and data sets is compared (SSL in the standard data set is equal
problems existed with one of the variables, VM2-VM3. to DM5-DM6 in the truss network data set). There are
This variable showed an average correlation with all other also significant differences among the data sets. P. setiferus
variables of 0.09. The remaining variables had average cor- and P. vannamel samples are larger in the standard data
relations ranging from 0.37 (DM4-DM5) to 0.69 (DM3- set, but P. slyhrostris are larger in the truss network data
VM4). This data set gave results from principal compo- set. Size differences among the species could impact the
nent analysis that were different from the standard data discriminant analysis. In this case, the size differences
set. PC 1 explained 60 To of the variation, PC 2 explains 9 among species are greater in the standard data set than
All loadings on PC 1 were between 0. 15 and 0. 23, except in the truss network data set and should contribute to
VM2-VM3 (0.02). Again this indicates that PC I is a size greater separation in the discriminant analysis of the stan-
component. The additional variables do not provide much dard data set. The discrimination is better however with
allometric variation to be explained by PC2. the truss network data set, which only serves to reinforce

68 NOAA Technical Report NMFS 92

Table 2
Comparison of variance components and heritabilities obtained from nested ANOVA using
simple variables and the first principal component scores. The eight simple variables were
selected to represent lengths, depths, and diagonals, as well as the range of heritability
estimates. Degrees of freedom in all cases: Family 7, Cage = 16, and Error = 106.

Variance Variance Variance
Variable (family) (cage) (error) Heritability

PCI 4.42 x 10' 0 1.04 x10' 0.60
DM5-VM5 1. 40 x 10-' 0 4.44 x 10-1 0.48
DM3-VM3 1.58 x 10-' 0.05 x 10-' 4.36 x 10' 0.53
DM2-VM2 1.92 x 10' 0.16 x 10-' 4.66 x 10' 0.57
DM5-DM6 1.86 x 10-' 0 4.85 x 10-1 0.55
DM5-DM4 2.50 x 10-' 0.36 x 10-' 5.74 x 10-' 0.58
DM3-DM2 1.72 x 10-' 0,35 x 10-' 5.56 x 10-1 0.45
DM4-VM3 2.08 x 10-3 0 4.56 x 10' 0.63
DM3-VM4 1.98 X 10-3 0.15 x 10-' 4.79 x 10' 0.57
Mean of 26 variables 0.49
SE 0.14

the conclusion that the truss network approach provides of 0. Clearly there are problems with this measurement.
additional information about species differences. DM2-VMl and VM2-VM3 have h2 values of only 0.28.
An average of all the h2 estimates is somewhat lower than
Multivariate Estimation of Heritability the one resulting from use of PCI scores.
Examination of the correlation matrix output from the
principal component analysis of the data from families Discussion
grown under experimental conditions showed that two of
the variables, VM2-VM3 and DMI-VMI were weakly Measurement problems are inherent in the development
correlated with the remainder of the data set. The correla- of a truss network design. In this study, they were reduced
tions of VM2-VM3 ranged from 0.25 (with DMI-VMI) by making changes in the landmarks used and by improv-
to 0.66 (with DM2-VM2). For DMINM1, the correla- ing photographic technique. Obtaining a simple set of land-
tions ranged from - 0.08 (with VM1-VM2) to 0.43 (with marks that remain homologous across development from
DM5-VM4). These ranges are distinctly lower than the postlarva to adult proved more difficult than expected. One
other variables and are indicative of measurement prob- source of error in the adult data set was the choice of points
lems. The juvenile shrimp correlations are higher than on movable bodyparts, (e.g., appendage insertions on the
those in the truss network data set from the adults. PC 1 adults). This problem was corrected by adopting new land-
explains 82 17o of the variance in the data set and has nearly marks on the juveniles that represented stable intersections
equal loadings of all of the variables. The coefficients were of body parts. In the case of the posterior rostral tooth,
between 0. 16 and 0. 23 which is indicative of their similarity juvenile shrimp often expressed different degrees of rostral
as estimators of size. tooth development from conspecifics, both adults and other
Results from multivariate and univariate scoring were juveniles. Thus slightly different landmarks had to be used
compared based on the heritabilities and variance com- to collect dimensional data from juvenile shrimp of 0. 2 to
ponents obtained from the same ANOVA design. Table 1.0 gm than were used for the adults. When working with
2 shows these results from comparison of eight univariate the juvenile shrimp, it was discovered that some points are
scores and the first principal component score. The results more difficult to see on a live shrimp than on a dead one,
show that a univariate approach can yield inaccurate com- (e.g., VM2). Better lighting and different photographic
parisons among genetic groups. Some variables have techniques were used to minimize this problem. The ex-
heritabilities that slightly exceed the estimation based on amination of different landmarks is continuing, especially
PC 1. However, most variables yield lower estimates of the for the juveniles.
family component of the variance in size, some drastically Several variables used in the truss network analysis ex-
lower. Using PC I scores in the ANOVA gives an F value hibited their unsuitability as size estimators by a reduced
for the model of 4.74 and a heritability (h2) of 0.60. The level of correlation with the other variables. The reason
most extreme case is VM1-VM2 which has an h2 estimate for low correlations from two of the variables, DM I NM I

Lester et al.: Image Analysis for Breeding Shrimp 69

and VM2-VM3, was related to difficulty in precisely
locating the landmark points, DMI, VMI, and VM2, on 8
the pictures of juveniles. Some covariation is expected 7
because all of the measurements are dimensions from the 6
same image and all will increase with size of the shrimp. 5
High correlation can be expected if all variables are reliable T_ 4 S
estimators of size and can be accurately measured. Vari- W 3
ables in the truss network data sets obtained from juveniles
FC 2
were all reasonable estimators of size as indicated by the < 1
high correlations and equality of loadings on PCI. > Y
_J 0
The problems associated with developing a truss network 4
approach are outweighed by the advantage obtained frorn 0 -1
Z
it. In this study, the three species were used to represent 0 .2 V
Z
any genetically distinct groups about which information -3
is needed on distinguishing characteristics. It was shown -4
that discrimination among Penaeus setiferus, Penaeus vannameil -5
and Penaeus st lirostris is possible using the standard data -61
set as seen in Figure 3. However, discrimination among -6 -5 -4 -3 -2 .1 0 1 2 3 4 5 6 7 8
these groups improved with the truss approach, as one can CANONICAL VARIATE 2
see by comparing Figures 3 and 4. It appears that the truss
network approach offers more information for the dis-
crimination of these genetic groups and the classification Figure 4
of individuals than the collection of standard measure- Plot of canonical variates I and 2 from discriminant analysis of
ments. Comparison of a data set of nine variables and one the truss network data set. The species are represented by ellipses
of 26 variables for discriminant analysis is biased against based on the 95 % confidence intervals of the species' centroids.
the data set with fewer variables. Although Figure 4 shows The ellipse for Penaeus setiferus is labeled S, Penaeus slyhrostris Y,
that classification of individuals by species would be easier and Penaeus vannamei V.
from a truss network approach, improved discrimination
among species could perhaps have been accomplished by
the addition of random variables to the standard data set. 0.45 (DM2-DM3) to 0.63 (DM4-VM3) is quite large for
Thus these results should not be taken as sufficient justifica- heritabilities of measurements taken from the same set of
tion for the truss network approach. A thorough justifica- individuals from the same set of families by the same
tion can be found in Bookstein et al. (1985) and is based observers. This variation is indicative of differences in
on information content and distribution within the data set - measurement error and information content among the
The use of principal component scores derived from the variables in the truss network. Principal component anal-
truss network data set has associated advantages and disad- ysis is one appropriate way to combine the information in
vantages. The number of variables obtained is much larger many size-related variables and minimize the effect of
than would normally be collected for a genetic analysis of measurement error. We believe that the use of multivariate
size or any other characteristic. If these variables were to classification analysis will provide a more reliable dis-
be measured by hand with calipers, the number of variables crimination based on size. We will continue to pursue this
would prove prohibitive. The major advantage of using approach in our experiments because it provides more ac-
multivariate classification scores for the selection process curate estimation of breeding value for size.
will be realized in the improved accuracy of estimating size It is presently impossible to employ the image analysis
and reduced probability of misclassification. While this ap- approach for selection of untagged shrimp because the
proach improves the analysis of size and shape of penaeid photography process and the computer analysis process are
shrimp for experimental purposes, programming a com- separated in time and space. Using our current digitizing
puter to locate homologous landmark points with accuracy process, it still takes the technician about two minutes to
is a major challenge. Some other approach to image anal- enter the landmark points from a single image. In addi-
ysis may be better for commercial application in selective tion, the time involved in taking good still photographs of
(D
@5D

breeding. living shrimp is considerable. These problems will be solved
There is considerable variation among the estimates of by the development of an integrated imaging and analysis
heritability obtained from univariate estimators of size. In system. The video image can be converted to a still image
some cases, it is not clear why more of the variance is par- by a frame grabber which operates in 1/30th of a second.
titioned into the family component, resulting in a higher The ability to locate and record landmark points or to
heritability. It is clear that the range seen in Table 2 of employ other image analysis techniques (e.g., definition

70 NOAA Technical Report NMFS 92

of the boundary of the shrimp and the area within that Benita Waas. Dr. Cecil Hallurn and Dr. Sharon Perkins
boundary) can be programmed into a fast microcomputer critically reviewed the manuscript.
which could obtain the required measurements. The results
of a preliminary classification analysis can be stored in the
computer and used to calculate the selection index. With Citations
a sufficiently powerful computer, the calculation of the in-
dex and comparison to a truncation value would take a Becker, W.A.
fraction of a second. Thus problems associated with the 1984. Manual of quantitative genetics, 4th ed. Academic Enter-
combination of many variables and too complicated an prises, Pullman, WA, 188 p.
Bookstein, F. L., B. Chernoff, R. L. Elder, J. M. Humphries Jr.,
analysis could be solved by integration of fairly simple hard- G.R. Smith, and R.E. Strauss.
ware and software. 1985. Morphometrics in evolutionary biology: The geometry of size
and shape change, with examples from fish. Acad. Nat. Sciences
Philadelphia, Spec. Publ. 15, 277 p.
Acknowledgments Falconer, D.S.
1981. Introduction to quantitative genetics, 2nd ed. Longman,
London, 340 p.
This research was sponsored in part by a grant to L.J. Lasley, J.F.
Lester from the Texas A&M University Sea Grant Col- 1978. Genetics of livestock improvement, 3rd ed. Prentice-Hall,
lege Program, supported by the National Oceanic and Englewood Cliffs, NJ. 492 p.
Atmospheric Administration, Office of Sea Grant, Depart- Lester, L.J.
ment of Commerce under grant number NA83AA-D- 1983. Developing a selective breeding program for penaeid shrimp
mariculture. Aquaculture 33:41-50.
00061. Support was also provided by the University of 1988. Differences in larval growth among families of Penaeus slyli-
Houston-Clear Lake. The assistance of the personnel of rostris and Penaeus vannamei. Aquacult. Fish. Manage. 19:243-25 1.
Laguna Madre Shrimp Farm, Marine Culture Enterprises, SAS Institute, Inc.
Agromarina de Panama, Granada Corp., and the Texas 1985. SAS User's Guide: Statistics, Version 5 ed. SAS Institute,
A&M Shrimp Mariculture Project in obtaining samples Inc., Cary, NC., 956 p.
Strauss, R.E., and F.L. Bookstein.
is gratefully acknowledged. Expert technical assistance was 1982. The truss: body form reconstructions in morphometrics.
provided by Margaret Dennison, Mary Byam-Smith, and Syst. Zool. 31:113-135.

Breeding Test on Abalone

HARUHIRO MOMMA

Hokkaido Hakodate Fisheries Experimental Station
Yunokawa Hakodate
Hokkaido 042, Japan

Introduction at maturity. Haliotis gigantea, for instance, is much larger
than H. discus (Inoue et al. 1985) because its growth pat-
Although production of artificial seed for abalone farming tern (Takayama 1940) and feeding behavior is different
in Japan has increased annually over the last 20 years, the (Momma 1980a). Thus, an increase in H. gigantea growth
total abalone harvest has continued to decline (Fig. 1) rate would need to be proportionally greater in order to
(Department of Statistics Information (DSI) 1967-86; have the same beneficial effect that a smaller increase would
Japan Sea-Farming Assoc. 1967-1985). In 1985 about 30 have on H. discus.
million seeds were produced. The proportion of recaptured The faster growing seedlings of a H. d. hannai culture
artificial seedlings to the total abalones harvested in plant- were recovered more frequently than the slow ones (Table
ing areas has usually been reported to be about 30-407c. 3). This higher rate of recovery illustrates that fast-growing
However, some of these reports have shown values ex- seed have higher survival rates than their slower growing
ceeding 9017o (Table 1). In these cases it is important to counterparts even though they were the same size at the
study the genetics of the seed abalone used. Table 2 shows time of release. It was previously reported by the author
some examples of recapture ratios of abalone seedlings. that juvenile abalone with initially rapid growth rates (until
Overall, the results suggest that seedling survival is usually a shell length of 16-17 mm is reached) maintain that trait
low and a large percentage of the seed suffers mortality.
Another reason to study their genetics is the fact that
differences in recapture ratios occur between species when
they are planted on the same fishing ground (Inoue et al.
1985). Are there differences in the quality of the recaptured
abalones compared to the seedlings which suffered mor-
tality? In order to increase the recapture ratio, it is impor-
tant to produce healthy seed abalone, improve seeding
methods, and control the factors which affect survival on 25
the planting ground.

20Z
C
Recapture Ratios and Seedling Quality a
X Cr
W 5.5
The survival rate within abalone species increases with seed C
0 0
size at the time of release. The extent of this increase, 15
however, varies between species in relation to their size CD
CL
5

.10
Figure 1 12 4.5
Total catch (0) and number of artificially produced abalone seed-
lings in Japan. Total catch: after fisheries statistics of Japan 5
1967-86. Department of Statistics Information, Ministry of
Agriculture, Forestry and Fisheries. Government of Japan. 4
Number of seedlings: after Materials on production, supply and
0
release of fingerlings for enhancement of fisheries resources in 1967 70 '75 so 45
Japan, 1967-1985, Japan SeamFarming Association.

72 NOAA Technical Report NMFS 92

Table 1
Proportion of artificial seed to abalone landed.

Proportion of
Prefecture Location seedlings (To) Species References

Hokkaido Toyoharna 85 H. d. hannai Momma (1986)
Iwate Toni 39 H. d. hannai Takeichi (1988)
Miyagi Utatsu 83 H. d. hannai Sasaki et al. (1987)
Fkushirna Nagasaki 59 H. d. hannai Sato et al. (1984)
Kanagawa Nagai 85 H. discus Tauchi (1984)
Fukuoka Oshima 54 H. discus FutaJima et al. (1985)

Table 2
Recapture rate of the planted seedlings.

Recapture
Prefecture Location rate (%) Species References

Hokkaido Shiribeshi 7 H. d. hannai Miyarnoto et al. (1982)
Iwate Toni 18 H. d. hannai Takeichi (1988)
Yarnagata Koiwagawa 26 H. d. hannai Ioka (1983)
Ibaragi Oarai 25 H. discus Kodarna (1985)
Kanagawa Jogashirna 26 H. gigantea Inoue (1965)

Table 3
Recapture rate of juvenile H. d. hannai in its nursery ground (from Momma et al. 1980).
Term of experiment was 116 days.

Average growth Released Number Recapture
Group ratio (Mm/d) size (mm) released Recovery rate M

Fast 71 13.8 � 1.3 964 170 17.6
Slow 28 13.9 � 1.9 185 20 10.8

throughout their early development for a period of at least method. This method has been repeated several times from
116 days (Fig. 2) (Momma 1980b). 1973 to 1989. The 6th generation of H. discus hannai pro-
In this paper the effects of artificial selection, h 'ybridiza- duced by the sib method of inbreeding did not result in
tion, and mutation on juvenile abalone growth are dis- higher than normal mortality, or growth acceleration (Fig.
cussed. The isolated fast-growth qualities were examined 3) (Momma 1987).
through various tests conducted under standardized rear- H. discus discus and H. discus hannaiwere introduced for
ing conditions with the following results. breeding by the usual method (Kikuchi and Uki 1971).
These seedlings were reared under the same conditions.
The growth and mortality for these species was compared.
Growth Rates Similar shell-growth rates were observed for these two
species, but mortality of the H. d. discus was higher than
The fastest growing individuals were selected from the off- H. d. hannal; therefore, the latter produced a greater com-
spring of one parent by mass selection when they were parative biomass. (Fig. 4) (Momma 1987).
150-200 days old: about 15 min in shell length. After these Hybridization studies were conducted on H. d. discus,
superior individuals were reared about 1 more year, they H. d. hannai, and H. kamischatkana using the usual method
were introduced for breeding and fertilized by the sib on the Ezo-abalone (H. d. hannai). In only one case did

Momma: Breeding Test on Abalone 73

150 -

o
ooo
o

MD
oo
120 oo1%o
-o o o
E o%
oooc
0-.. %
0.
0
0 0
0
00
d) 90 00 0
E

CL
X

so -
M

E
=L

to 30

0
Figure 2
The relationship between thejuvenile stage (shell length:16-17
0 40 60 80 100 120 mm), the growth ratio (pm/d), and the growth ratio during the
Juvenile stage (till shell length: 16-17mm) experimental term (116 d). 0 = fast-growth group at 153-269
growth ratio (pm/day) d old and 0 the slow-growth group at 570-686 d old (Mom-
ma 1980b).

30-
- G2 = 30 June 1986
- G6 = 7 July 1986
E
E 20m

50 100 150 200 250 300
Age (days)

Figure 3
Growth lines of H. discus hannai seedlings; 0 and 0 = the mean shell length of 2nd (G2) and
6th generation (G6), respectively. (G2 fertilized 30 June 1986; G6 fertilized 7 July 1986) (Momma
1987).

74 NOAA Technical Report NMFS 92

4 -100

3 75

- hannai , iomass
- hannai :urvival ratio

(A 2 - discus Biomass -50 <
E discus Survival ratio
0

204 4" 6H M inIll 1260
A4P (days)

Figure 4
Biomass and survival rate of H. d. hannai and H. d. discus. (Momma 1987).

Table 4
List of the results from analyzed enzymes. (from Fujio 1984; Fujino 1979). D = digestive
diverticula, M = shell-muscle, P polymorphic, P* polymorphic less than 517o, M
monomorphic.

Enzyme Locus Tissue Polymorphism

Acid phosphatase Acp D P.
Adenylate Kinase Ak M M
Aspartrte arninotransferse Aat M P
Esterase Est-I D P
-2 D M
-3 D M
a-glycerophosphate Dehydrogenase aGpd M M
Glucosephosphate Isornerase Gpi M M
Isocitrate Dehydrogenase Idh-I M M
Lactate Dehydrogenase Ldh-I M p
-2 M M
Leucin Aminopeptidase Lap- I D P
-2 D M
Malate Dehydrogenase Mdh-I M M
-2 M P*
Malic Enzyme Me M M
Mannosephosphate Isomerase Mpi M P*
Octanol Dehydrogenase Odh D P
6-phosphogluconate Dehydrogenase 6Pgd M P
Phosphoglucoselsomerase Pgi D P*
Phosphoglucornutase M-1 M P
-2 M P
Superoxide Dismutase Sod M M
Tetrazolium Oxidase To-2 D P*
-3 D P.

Water temperature (OC)

bl)

co
7S 0)

o M
q*
OD
0)
co
>,
IM

co
00

04 cr) cf)

(Luw) LIIBUGI lia4s
(Luw) LIIBUGI 11GLIS
75

76 NOAA Technical Report NMFS 92

the hybrid of H. d. hannal and H. kamtschalkana show im- Ioka, 1.
proved growth rate in cold water. (Fig. 5) (Momma 1988). 1983. Yarnagataken ni okeru horyu ezo-awabi no seicho to seizan-
ritu no sisanrei. Otsuchi Marine Research Center Report 9:
Triploids produced by interference of polar body pro- 52-51.
jection (both lst and 2nd polar body extrusions) using the Japan Sea-Fasming Association.
method of Arai et al. (1982) did not differ in growth from 1967-1985. SaibaiGyogyoSyubyoSeisan,NyusyuHoryujisseki
diploid seed (Fig. 6) (Momma 1988). Showa 43-52 Nendo. (Materials on production, supply, and release
of fingerlings for enhancement of fisheries resources in Japan.)
Japan Sea-Farming Assoc., Japan.
Kikuchi, S., and N. Uki.
Abalone Breeding 1974. Technical study on artificial spawning of abalone, genus
Haliods-11. Effect of irradiated sea water with ultraviolet rays
Enzyme polymorphism of this species (Table 4) (Fujio on inducing to spawn. Bull. Tohoku Reg. Fish. Res. Lab. 33:
79-86.
1984; Fujino 1979), and the chromosome handling tech- Kodama, S.
niques and growth characteristics of other species have been 1985. Awabi no horyu koka ni tuite. Bull. Japanese Soc. Fish.
reported. Above all, based on the study of wild abalone Oceanog. 47/48:139-142.
populations, as well as the relationship of age to fitness Miyamoto, T., K. Saito, M. Ito, and Y. Mizutori.
measured at the esterase M locus, the deficient animals 1982. On the releasing of the cultured seeds of Ezo-abalone, Haliods
discus hannai, in the northern coast of Siribesi, hokkaido, Hokusui-
reveal more homozygosity (Fujino 1978). Population shi-geppo. Hokkaido Fish. Exp. Sm. 39(8):169-208.
genetics and thremmatological studies may contribute con- Momma, H.
siderably to the abalone industry. 1980a. On the abalone behavior under the influence of predator.
Michurian Seibutsugaku kenkyu 16(l):60-69.
1980b. Studies on the variation of the abalone-1. On the growth
of the different aged young abalone. Aquaculture, 28(3):142-146.
Citations 1986. Awabijinkoshubyuhouryujigyonokokachosa. Sodateru-
gyogyo 159:2-4.
Arai, K., H. Tsubaki, and K. Fujino. 1987. Ezo-Awabi. Annu. Rep. Hokkaido Inst. Maricult. 61:
1982. Chromosomes of Haliotis discus hannai Ino and H. discus Reeve. 53-61.
Bull. Jap. Soc. Sci. Fish. 48:1689-1692. 1988. Kairui shubyo chukan ikusei siken. Ezo-Awabi. Annu.
Department of Statistics Information (DSI). Rep. Hokkaido Inst. Maricult. 62. 57-59.
1967-1986. Fishery statistics of Japan 1967-86. DSI,Jpn.Min. Momma, H., K. Kobayashi, T. Kato, Y. Sasaki, T. Sakamoto, and
Agric., Forest. Fish., Tokyo. (In Japanese.) H. Murata.
Fujino, K. 1980. On the artificial propagation method of abalone and its ef-
1978. Genetic studies on the Pacific Abalone-1. Excessive Homo- fects on rocky shore-1. Remaining ratio of the artificial seed
zygosity in Deficient Animals. Bull. Jpn. Soc. Sci. Fish., 44(7), abalones (Haliotis discus hannai INO) on the lattices artificial reefs.
767-770. Aquaculture, 27(4):212-216.
1979. Apparent physical deficiency associated with inbreeding struc- Sasaki, R., K. Takahasi, T. Kawamura, and Y. Tasiro.
ture suggested by biochemical polymorphism in the Pacific 1987. Miyagiken hokubu kaiku ni okeru horyu ezo-Awabi no chosa
Abalone. In Proc. 16th int. conf. on animal blood group and jirei. Tohoku Reg. Fish. Lab., Report 61. 15-29.
biochemical polymorphism: Leningrad, USSR (16 August 1978) Sato, M., J. Owada, and H. Suzuki.
p. 245-256. USSR Organizing Comm. Int. Soc. Animal Blood 1984. Jinko shubyo awabi, uni horyu koka chosa. Annu. Rep.
Group Res. Fukushima Pref. Fish. Exp. Stn:181-190.
Fujio, Y. Takayama, K.
1984. Studies on genetic analysis of fishies by isozyme methods. 1940. On the growth rate of the Abalone at Mie prefecture. Suisan
Minist. Agric. Forestry and Fisheries, Government ofJapan, 65 p. kenkyushi (precedes Bull. Jpn. Soc. Sci. Fish.) 35(4):99-100.
Futajima, K., T. Ito, and G. Kishimoto. Takeichi, M.
1985. Awabi no saibai gyogyokanri ni kansuru kenkyu-1. Annu. 1988. Tairyo horyu sareta ezo-awabi jinko shubyo no kaishuritu
Rep. Fukuoka Pref. Fish. Exp. Stn, p. 239-257. to seizanritu Saibai Gyogyo Gijitu Kaihatu Kenkyu (Journal on
Inoue, T. technical development for artificial enhancement of fisheries
1965. Hyoshiki horyu kara mita Awabi zoku no ishoku ni kansuru resources). Saibaigiken 17(l):27-36.
1. 2. Suisan-Zoshoku. Ext. 5:23-31. Tauchi, M.
Inoue, T., M. Tauchi, and M. Chikayama. 1984. Growth, age composition, survival rate and recapture rate
1985. Diagnosis of species participate in effect on artificial seeds of Japanise black abalone, Haliods discus Reeve, at Nagai. Sci.
release of abalone, Haliods spp. Aquaculture, 32(4), 193-198. Rep. Kanagawa Fish. Exp. Sm. 6:17-21.

Two-Stage Hybridization and Introgression
for Improving Production Traits of Red Tilapias

LESLIE L. BEHRENDS, JOHN B. KINGSLEY, and ALBERT H. PRICE III

Agriculture Research Department
Tennessee Valley Authority
Muscle Shoals, AL 35660-1010

ABSTRACT

Two-stage hybridization and introgression were evaluated as breeding plans to develop im-
proved hybrid populations of red tilapia. Both plans were successful in overcoming interspecific
breeding barriers, thus allowing development of hybrid populations (both red- and non-nal-colored
phenotypes), with a mix of desirable production traits. The hybrids were evaluated in replicated
polyculture experiments involving blue tilapia, Tilapia aurea, channel catfish, Ictalurus punctatus,
freshwater prawn, Macrobrachium rosenbergii, and Asiatic carps. Yield trials were conducted in earthen
ponds. Two-stage hybridization ( [ T aurea x red tilapia] x [red tilapia x T. nilotica ] ) was used
to produce red- and normal-colored hybrids that grew to average weights of 313 and 292 grams
respectively in 133 days. Under similar conditions (communal stocking at equal density), T aurea
controls averaged 2 52 grams. Introgressive breeding techniques ff T aurea x red tilapia] followed
by two generations of backcrossing red males [F- I and F-2 generations respectively], to female
T. aurea) were used to develop a cold tolerant tilapia hybrid with both red- and normal-colored
phenotypes. After 146 days of culture, both the red- and normal-colored hybrids averaged 296
grams, while T. aurea controls averaged 276 grams. Irrespective of breeding plan, mortality of
the red phenotype was significantly greater than either of the normal-colored phenotypes.

Introduction stains have not been well documented (Sipe 1979; Fitz-
gerald 1979). However, morphometric and electrophoretic
Individual (mass) selection for rapid growth in the mater- data indicate common ancestry with a red-mutant strain
nal mouthbrooding genus Tilapia has had limited success, of T mossambica (Galman and Avtalion 1983; Halstrom.
indicating that the fraction of additive genetic variation is 1984). The widely distributed U.S. "Florida strain,"
small relative to total genetic variation (Chan May Tchien derived from the cross female T hornorum x male T mos-
197 1; Tiechert-Coddington 1983; Hulata et al. 1986). In sambica (red mutant), has production traits similar to its
such instances, hybridization and introgression can be parental lines: early sexual maturation, slow growth, and
utilized to avail for nonadditive sources of genetic varia- lack of cold tolerance (Halstrom. 1984).
tion (Brody et al. 1980). In replicated yield trials, Florida strain young-of-the-year
Within the interfertile genus Tilapia, there are species fingerlings grew only 50 % as fast as blue tilapia, T aurea,
which exhibit fast growth (T aurea; T nilotica), high fecun- (Behrends et al. 1982). Thus, it was hypothesized that pro-
dity (T mossambica), cold tolerance (T aurea), and salinity duction traits could be improved by hybridizing the Florida
tolerance (T. mossambica; T aurea). Also, red-colored mu- strain with T aurea or T n1lotica, two of the faster-growing
tant strains of T. mossambica and T nilotica have arisen and later-maturing species. Subsequently, Behrends and
spontaneously from normal-colored populations (Fitzgerald Smitherman (1984) reported that cold tolerance and the
1979; McAndrew et al. 1988). Thus, a diversity of genetic red color trait could be incorporated into a single popula-
resources is available for developing hybrids or synthetic tion via hybridization (female T aurea x male Florida
breeds in which favorable traits are combined. strain), followed by recurrent backcrossing of red hybrid
Several hybrid strains of red tilapia have been com- males to T aurea females.
mercially produced and widely disseminated throughout This paper will present results of studies designed to
Southeast Asia and the Western Hemisphere during the evaluate the use of two-stage hybridization and hybridiza-
past fifteen years. Breeding plans used in developing these tion followed by repeated backcrossing (introgression)

78 NOAA Technical Report NMFS 92
-Breeding Plan 1983 Breeding Plan 1984
Hybridization Introgression
?Tilapia aurea ?Red tilapia Tilapia aure x dRed tilapia
X (Florida strainY W (Florida strainY
d Red tilapia X F-1**
(Florida strainY dTilapia nilotica (25 50% red; 50% normal)
I 1 1%
F- 1 hybrid F- 1 hybrid ?T. aurea x (f F-1 (red)
(@_-50% red; 0_--50% red; I
50% normal) 50% normal) F-1 backcross
It fr (=-50% red; 50% normal)
?F-1 (red) x dF-1 (red) 1%
T. aurea x d F- 1 backcross
J1 I (red)
FTrihyb IF-2 back

*Florida strain of red tilapia derived from *Florida strain of red tilapia derived from
9 T. hornorum x o4 T. mossambica (red mutant). VT. hornoru x d4T. mosambica (red mutant).
"Red phenotypes are heterozygous for red "Red broodstock were heterozygous for
color. Normal phenotypes are homozygous the red color trait, thus approximately
for normal color. Pink phenotypes are 50% of progeny were red and 50% normal.
homozygous for pink color.

Figure I Figure 2
Breeding plan for developing cold-tolerant red tilapia hybrids. Breeding plan for developing fast growing red tilapia hybrids.

for developing red tilapia hybrids with improved culture channel catfish (Ictalurus punctatus, 10000/ha), freshwater
performance. prawn (Macrobrachium rosenbergii, 20000/ha), grass carp
(Clenopharyngodon idella, 120/ha) and F-1 hybrid Asian carp
(An'stichthys nobilis x Hypopthalmichthys molitrix, 160-500/ha).
Materials and Methods Fingerlings developed via two-stage hybridization were
stocked in 1983, while introgressed hybrids were stocked
Young-of-the-year red- and normal-colored (grey) tilapia in 1984. Red- and normal-colored hybrids and blue tilapia
fingerlings were produced at the Tennessee Valley Author- were costocked into each pond at equal densities (800/ha)
ity's Research Farm from 1981 to 1984 using breeding for a total density of 2 400/ha (Tables 1 and 2). Blue tilapia
plans illustrated in Figures 1 and 2. In both trials, red served as within- and between-years controls. Costocking
broodstock heterozygous for the red color trait were used of different genetic groups, also referred to as communal
to ensure production of both red- and normal-colored testing, has been shown to be an efficient method for
phenotypes. By comparing the two hybrid phenotypes, it evaluating various tilapia strains and hybrids under con-
was possible to assess the effect of the red gene complex ditions of common environment (Wohlfarth et al. 1983).
on growth, yield, and mortality. Throughout each study, a 32 % crude-protein floating
In each of two years (1983 and 1984), three genetically catfish ration was broadcast onto the pond surfaces daily.
and phenotypically distinct tilapia populations (Tables I Feeding rates were adjusted daily based on projected cat-
and 2), were costocked into four 0.05-ha ponds along with fish biomass (Behrends et al. 1985). After 130 to 140 days

Behrends et al.: Hybridization and Introgression of Red Tilapias 79

Table I
Stocking and harvest data for three tilapia populations costocked into 0.05-ha earthen ponds.
Polyculture yield trial 1983, Tennessee Valley Authority, Muscle Shoals, Alabama. Each mean
represents the average of four ponds.'

Two-stage hybridization

Red hybrid Normal hybrid Tilapia aurea

Stocking density (#/ha) 800 800 800
Culture duration (days) 133 133 133
Initial weight (g/fish) 2.0 2.1 2.3
Final weight (g/fish) 313 a' 292 b 252 c
Males (mean � SD) 330 � 63.0 318 � 59.0 283 � 38.0
Females (mean � SD) 256 � 58.4 254 � 57.0 225 � 35.3
Sex ratio (To males) 77 59 48
Daily gain (g/fish/d)
Males (mean) 2.5 2.4 2.1
Females (mean 1.9 1.9 IJ
Survival (To) 88 93 100
Yield (kg/ha) 219 a 216 a 200 b Total 635 kg/ha

'Tilapias were stocked into polyculture ponds containing channel.catfish (10000/ha) freshwater prawn
(20000/ha) and Asiatic carps (300-600/ha) (Behrends et al. 1985).
Means followed by different letters are significantly different (P<0.05, Student- Newman-Kuels test).
Horizontal comparisons only.

Table 2
Stocking and harvest data for three tilapia populations costocked into 0.05-ha earthen ponds.
Polyculture yield trial 1984, Tennessee Valley Authority, Muscle Shoals, Alabama. Each mean
represents the average of four ponds . a

Introgressed hybrids

Red hybrid Normal hybrid T21apia aurea

Stocking density (#/ha) 800 800 800
Culture duration (days) 146 146 146
Initial weight (g/fish) . 0.6 0.9 0.9
Final weight (g/fish) 296 a' 296 a 279 b
Males (mean � SD) 299 � 65.2 325 � 65.4 298 � 63.1
Fema 'les (mean SD) 268 � 44.9 260 � 62.0 254 � 46.1
Sex ratio (To males) 90 56 49
Daily gain (g/fish/day)
Males (mean) @2. 0 2.2 2.0
Females (mean) 1.8 1.8
Survival 65 92 71
Yield (kg/ha) 153 b 217 a 156 b Total 526 kg/ha

'Tilapias were stocked into polyculture, ponds containing channel catfish (10000/ha) freshwater prawn
(20000/ha) and Asiatic carps (300-600/ha) (Behrends et al. 1985).
'Means followed by different letters are significantly different (P<0.05, Student-Newman-Kuels test). Horizon-
tal comparisons only.

of culture (May-October), ponds were drained and har- during 1983 and 1984 included effects due to phenotype,
vested. Tilapias within ponds were sorted by phenotype pond nestedwithin phenotype, sex, and all first-order in-
and sex within phenotype. Individuals were then weighed teractions. Student-Newman-Kuel's Multiple Comparisons
to the nearest gram. Test was used to separate means (within years), at a sig-
Statistical models used to evaluate growth and yield data nificance level of P<0.05 (Barr et al. 1979). Details con-

80 NOAA Technical Report NMFS 92

cerning culture and yields of catfish, prawns, and carps to faster growth and excellent survival. Mortality of
are presented in a companion paper (Behrends et al. 1985). normal-colored hybrids averaged only 87o, while mortal-
ity rates for red hybrids and T aurea averaged 35 and 29 7o
respectively (Table 2). Sex ratio of the red hybrid popula-
Results - tion was highly skewed towards males (907o), while the
percentage of males in the normal-colored hybrid and
Two-Stage Hybridiation T aurea populations was 56 and 497o respectively.
Two-stage hybridization (Fig. 1), was an effective breeding
strategy for enhancing growth rate and yield of red- and
normal-colored tilapia hybrids (Table 1). After 133 days Discussion
of culture, mean harvest weights (sexes pooled) of red and
normal phenotypes were 313 and 292 g respectively. Red Both two-stage hybridization and introgression were effec-
hybrids were heavier at harvest (P<0.05) than their tive breeding strategies for enhancing growth rates of red-
normal-colored counterparts, indicating a positive growth and normal-colored hybrids. However, relative to T aurea
effect of the red gene complex. In comparison, T aurea con- controls, two-stage hybridization improved growth rates
trols averaged 252 g, and weighed less (P<0.05) than either and yields to a greater extent than did introgression. This
of the hybrids. Irrespective of phenotype, males were 20 is not surprising because introgression, as practiced, is
to 307o heavier than females (P<0.05). Relative ranking analogous to inbreeding; heterozygous hybrid populations
of mean harvest weights by phenotype were the same in become progressively more homozygous with repeated
all ponds, indicating that pond (environment) x phenotype generations of backcrossing to a small founder population.
(genotype) interactions were not a significant source of Introgression is recommended for incorporating a desirable
variation. dominant trait from one population into a second, more
Net production of the three phenotypes ranged from 200 productive population (Kirpichnikov 1981).
to 219 kg/ha for a combined yield of 634 kg/ha (Table 1). During both years, mortality figures for red phenotypes
Mortality of red hybrids was consistently greater than mor- were higher than rates for normal-colored phenotypes.
tality of normal-colored hybrids and T aurea. Sex ratios Elevated levels of mortality have also been reported in other
(male:female) of red and normal-colored hybrids averaged production studies (Behrends et al. 1982, 1988), and in
7 7 % and 59 01o males respectively, while the T aurea pop- hatchery research (El-Gamal et al. 1988). This indicates
ulation averaged 48% males: near the expected 1:1 sex that the red gene complex, while enhancing growth, may
ratio. have had a negative pleiotropic effect on viability. Color
Despite their higher rates of mortality, hybrids yielded mutations in green sunfish, Lepomis cyanellus, and common
7-10% greater than T aurea (P<0.05). Improved yields carp, Cyprinus carplo, have resulted in reductions in viabil-
were due to improved growth of both males and females ity, growth or both (Dunham and Childers 1990; Kir-
and male-dominated sex ratios. pichnikov 198 1). Other red mutant strains of tilapia have
been identified (McAndrew et al. 1988; Tave 1989), and
Introgression should be evaluated with respect to growth rate, viability,
and use in hybrid breeding programs.
Introgressive breeding (Fig. 2), similarly improved growth Breeding plans in this study worked well for combining
performance of red- and normal-colored hybrids (Table traits of closely related species. While many species of
2), although not to the extent of two-stage hybridization. maternal mouthbrooding tilapia are interfertile, commer-
After 146 days of culture, mean harvest weights of both cial production of most F-1 hybrids is difficult owing to
red- and normal-colored phenotypes (sexes pooled) were species- specific differences in breeding behavior (Lee 1979;
296 and 296 g respectively. This differs somewhat from Hulata et al. 1985). In the present studies, problems related
results of the 1983 study, where the red hybrid grew to breeding behavior were overcome by resorting to double
faster than it's normal-colored counterpart. At harvest, hybridization or introgression. In both breeding plans, each
T. aurea controls averaged 279 g and weighed less than of the respective parents (F-1 and subsequent backcross
either hybrid (P<0.05). Irrespective of phenotype, males generations), had a complete or nearly complete set of
were 15-20% heavier than females (P<0.05). As in the chromosomes in common. This condition was apparently
previous years study, relative rankings of mean harvest sufficient to overcome natural reproductive isolating
weights (by phenotype) were the same between the four mechanisms which can impede reproduction between
replicate ponds. closely-related species. Similar interspecific breeding plans
Net production of the three phenotypes ranged from 153 may be practical for developing unique hybrids or synthetic
to 217 kg/ha, with a combined yield of 526 kg/ha (Table strains for special environments, for instance, a fast grow-
2). Mean yield of the normal-colored hybrid was greater ing, salt-tolerant red hybrid for net-pen culture in tropical
(P< 0. 05) than yields of the red hybrid and T aurea owing estuaries.

Behrends et al.: Hybridization and Introgression of Red Tilapias 81

Citations Fitzgerald, W.J.
1979. The red-orange tilapia. Fish Farming Int. 6(l):26-27.
Barr Aj., J.H. Goodnight, J.P. Sall, W.H. Blair, D.M. Chilko, Galman, 0., and R.R. Avtalion.
K.A. Council, and JT. Helwig. 1983. A preliminary investigation of the characteristics of red tdapias
1979. SAS User's Guide. SAS Institute Inc., Raleigh, NC, from the Philippines and Taiwan. In Int. symposium on tilapia
494 p. in aquaculture; 8-13 May 1983, Naareth, Israel (L. Fishelson and
Behrends, L.L., and R.O. Smitherman. Z. Yaron, Compilers), p. 291-301. Tel Aviv Univ., Tel Aviv,
1984. Development of a cold tolerant population of red tilapia Israel,
through introgressive hybridization. J. World Maricult. Soc. Halstrom, M.L.
15:172-178. 1984. Genetic studies of a commercial strain of red tilapia. M.S.
Behrends, L.L., R.G. Nelson, R.O. Smitherman, and N.M. Stone. Thesis, Auburn Univ., Auburn, AL, 80 p.
1982. Breeding and culture of red-gold color phase of tilapia. J. Hulata, G., S. Rotbbard, J. Itzkovich, G. Woh1farth, and A. Halevy.
World Maricult. Soc. 13:210-220. 1985. Differences in hybrid fry production between two strains of
Behrends, L.L., J.B. Kingsley, and A.H. Price III. nile tilapia. Prog. Fish Cult. 47(l):42-49.
1985. Polyculture of freshwater prawns, tilapia, channel catfish, Hulata, G., G. Wohlfarth, and A. Halevy.
and Chinese carps. J. World Maricult. Soc. 16:437-450. 1986. Mass selection for growth rate in the Nile tilapia (Oreochromis
1988. Bidirectional backcross selection for body weight in red nilolicus). Aquaculture 57:177-14.
tilapia. In The second international symposium on tilapia in Kirpichnikov, V.S.
aquaculture; 16-20 March 1987, Bangkok, Thailand (R.S.V. 1981. Genetic Bases of Fish Selection. Springer-Verlag, New
Pullin, T. Bhukaswan, K. Tonguthai, and J. L. Maclean, eds.), York, NY, 410 p.
p. 125-133. Dep. Fisheries, Bangkok, Thailand, and Int. Center Lee, J.C.
for Living Aquatic Resources Manage., Manila, Philippines, 1979. Reproduction and hybridization of three cichlid fishes, Tilapia
ICLARM Conf. Proc. 13. aurea, T hornorum, and T nilotica in aquaria and plastic pools.
Brody, T., N., Storch, D. Kirst, G. Hulata, G. Wohlfarth, and R. Moav. Ph.D. Diss., Auburn Univ., Auburn, AL, 84 p.
1980. Application of electrophoretic genetic markers to fish breed- McAndrew, Bj., F.R. Roubal, Rj. Roberts, A.M. Bullock, and
ing. III: Diallel analysis of growth rate in carp. Aquaculture I.M. McEwen.
20:371-379. 1988. The genetics and histology of red, blond and associated color
Chan May Tchien. variants in Oreochromis niloticus. Genetica 76:127-137.
1971. Experience with the determination of realized weight Sipe, M.
heritability in the tilapia, Tilapia mossambica. Genetica 7(t): 1979. Golden Perch. Commercial Fish Farmer & Aquaculture
53-59. News 3(5):56,
Dunham, R.A., and W.F. Childers. Tave, D., M. Rezk, and R.O. Smitherman.
1980. Genetics and implications of the golden color morph in green 1989. Genetics of body color in Tilapia mossambica. J. World
sunfish. Prog, Fish Cult. 42(3):160-163. Aquacult. Society 20(4):214-222.
El Carnal, A.A., R.O. Smitherman, and L.L. Behrends. Teichert-Coddington, D.
1988. Viability of red and normal-colored Oreochromis aureus and 1983. Bidirectional mass selection for growth rate in Tilapia
0. niloticus hybrids. In The second international symposium on nilotica. M.S. Thesis, Auburn Univ., Auburn, AL, 48 p.
tilapia in aquaculture; 16-20 March 1988, Bangkok, Thailand Wohlfarth, G. W., G. Hulata, S. Rothbard, J. Itzkovich, and A. Halevy.
(R.S.V. Pullin, T. Bhukaswan, K. Tonguthai, andj.L. Maclean, 1983. Comparisons between interspecific hybrids for some produc-
eds.), p. 153-157. Dep. Fisheries, Bangkok, Thailand, and Int. tion traits. In International symposium on tilapia in aquaculture
Center for Living Aquatic Resources Manage., Manila, Philip- (L. Fishelson and Z. Yaron, Compilers), p. 559-569. Tel Aviv
pines, ICLARM Conf. Proc. 15. Univ., Tel Aviv, Israel.

Guidelines for Contributors

NOAA Technical Report NMFS

PREPARATION umn headings; do not deviate from the unit of measure within a column.
A manuscript should follow the sequence outlined below and include the Table footnotes should be noted consecutively in italic letters across the
following information. page from left to right and then down. Use asterisks, however, to mark
significance in statistical data. Since all tables are typeset, they need not
TitlePage Include authors' full names and complete mailing addresses. be submitted camera-ready.
Contents Page Outline the major divisions and subdivisions of the Figures Photographs and line drawings should be of professional
manuscript. quality-clear an *d concise-and reducible to 42 picas for full-page width
or to 20 picas for a single-column width, and a maximum of 54 picas
Abstract One paragraph and not to exceed one double-spaced typed high. All graphic elements in illustrations must be proportioned to in-
page. Should include a sentence or two explaining to the general reader sure legibility when reduced to fit the page format. Line weight and let-
why the research was undertaken and why the results should be viewed tering should be sharp and even. Lettering should be upper and lower
as important. Abstract should convey the main point of the paper and case, and vertical lettering should be avoided whenever possible (except
outline the results or conclusions. No footnotes or references. for vertical, y, axis). Zeros should precede all decimal points for values
less than one. Reproductions of line artwork are accepted in the form
Text A brief introduction should portray the broad significance of the of high-quality photographic prints from negatives or photomechanical
paper. The entire text should be intelligible to readers from different transfers (PMTs). Halftones should be sharply focused with good con-
disciplines. All technical terms should be defined, Follow the US. Govern- trast. Micron rules should be inserted on electron micrographs, even when
ment Printing Office Style Manual, 1984 ed., and the CBE Style Manual, 5th ed. magnification is included in the figure legend. There should be a clear
Fishery nomenclature should follow the American Fisheries Society Special distinction between identifying letters (press-on or overlays) and the back-
Publication 12, A List of Common and Scientific Names of Fishesfrom the United ground of photograph. Label each figure in pencil on the back. Send only
States and Canada, 4th ed. xerox copies of figures to the Scientific Editor; originals or photographic
prints will be requested later when the manuscript is accepted for
Define all abbreviations, acronyms, and symbols in text, equations, and publication.
formulae. Abbreviate units of measure only when used with numerals
or in tables and figures to conserve space. The numeral one (1) should Copies of Published Reports For single monographs, the senior author
be typed as a one, not a lower-case el (1). and his/her organization each receive 50 copies free-of-charge. If the
Measurements should be expressed in metric units. Other equivalent units monograph forms part of a proceedings or collection of articles, the senior
may be given in parentheses. author and his/her organization each receive 50 copies of the monograph
and one copy of the collective work. Additional copies may be purchased
Text footnotes should be numbered in Arabic numerals and typed on a in lots of 100.
separate sheet from the text. Footnotes are not used for reference material
nor for personal communications, but rather for explanations and defini-
tions in the text and for contribution numbers. SUBMISSION

Personal communications are noted parenthetically in the text (full name, Send manuscript (original plus two copies) to the Scientific Editor:
affiliation, brief address including zip code, month and year). Dr. Linda L. Jones, Scientific Editor
Acknowledgments Gather all acknowledgments into a brief statement National Marine Mammal Laboratory, F/AKC3
at the end of the text. Give credit only for exceptional contributions and National Marine Fisheries Service, NOAA
not to those whose contributions are part of their normal duties. 7600 Sand Point Way, N.E.
Seattle, WA 98115-0070
Citations All written sources should be listed in the Citations section, Once the manuscript has been accepted, a software copy, following the
including unpublished and processed material. In text, cite as Smith and disk specificiations described below, should be sent to the Managing Editor:
Jones (1977) or (Smith and Jones 1977); if more than one citation, list
chronologically (Smith 1936; Jones 1975; Doe 1977). All sources cited in Sharyn Matriotti, Managing Editor
the text should be listed alphabetically by the senior authors' surname NOAA Technical Reports NMFS
under the heading CITATIONS. Abbreviations of periodicals and serials National Marine Fisheries Service, NOAA
TM
should conform to the Biosis Data Base . Indicate whether sources are Scientific Publications Office, F/NWRI
in a language other than English. For informal literature, include address 7600 Sand Point Way, N.E.
of author or publisher. Authors are responsible for the accuracy of all Seattle, WA 98115-0070
citations.
Disk Specifications A manuscript should be submitted in ASCII for-
Tables should supplement, not duplicate, the text. Each table should be mat (i.e., in a MS-DOS "print" or "nondocument" file) and should
numbered and cited consecutively, with headings short but amply descrip- be placed on a 5 Y2 -inch (preferably) or 3 Y+ -inch disk that is double-sided,
tive so that the reader need not refer to the text. For values less than one, double or high density and that is compatible with either IBM or
zeros should precede all decimal points. Indicate units of measure in col- Apple/Macintosh systems.

UNITED STATES
DEPARTMENT OF COMMERCE BULK RATE
NATIONAL OCEANIC AND ATMOSPHERIC
ADMINISTRATION POSTAGE & FEES PAID
NATIONAL MARINE FISHERIES SERVICE
SCIENTIFIC PUBLICATIONS OFFICE U.S. Department of Commerce
BIN C15700 Permit No. G-19
SEATTLE, WA 98115
OFFICIAL BUSINESS
Penalty for Private Use, $300

NOAA SCIENTIFIC AND TECHNICAL PUBLICATIONS

7he National Oceanic and Atmospheric Administration was established as part of the Department of Commerce
on October 13, 1970. The mission responsibilities of NOAA are to assess the socioeconomic impact of natural and
technological changes in the environment and to monitor and predict the state of the solid Earth, the oceans and their
living resources, the atmosphere, and the space environment of the Earth.

The major components of NOAA regularly produce various types of scientific and technical information in the
following kinds of publications:

PROFESSIONAL PAPERS-Important definitive research TECHNICAL SERVICE PUBLICATIONS-Reports con-
results, major techniques, and special investigations. taining data, observations, instructions, etc. A partial
listing includes data serials; predictions and outlook
CONTRACT AND GRANT REPOR`rS-Reports prepared periodicals; technical manuals, training papers, planning
by contractors or grantees under NOAA sponsorship. reports, and information serials; and miscellaneous
technical publications.
ATLAS-Presentation of analyzed data generally in the TECHNICAL REPORTS-Jourrial quality with extensive
form of maps showing distribution of rainfall, chemical details, mathematical developments, or data listings.
and physical conditions of oceans and atmosphere, distribu- TECHNICAL MEMORANDUMS-Reports of prelim-
tion of fishes and marine mammals, ionospheric condi- inary, partial, or negative research or technology results,
tions, etc. interim instructions, and the like.

Information on availability ofNOAA publications can be obtainedfrom:

U.S. Department of Commerce
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161

3 6668 00003 6147