quantifying nutrient requirements of fishl

Upload: giuseppe-av

Post on 02-Jun-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/11/2019 Quantifying Nutrient Requirements of Fishl

    1/6

    Quantifying

    Nutrient Requirements

    of

    Fishl

    IsRAHtN.{ H.

    ZnrrouN,

    DUINIE

    E. Ur-lREv,

    WIr-r-Ierr,t

    T.

    MacE.e

    Department of Animal Husbandry, Michigan State University, East

    Lansing, Mich.48824,

    USA

    J o H u

    L . G t l r

    Department

    of

    Dairy

    Science,

    Michigan State University,

    East Lansing,

    Mich.48824,

    US A

    aNo WsnNen

    G.

    BEncEN

    Department

    of Animal

    Husbandry,

    Michigan State

    University, East

    Lansing, Mich.48824,

    USA

    ZetrouN,

    L H., D. E. Ur-r-ney,W.

    T. Me.cEE,

    .

    L.

    GrLt- , .cNo

    W. G.

    BBncr.N. 1976.

    Quanti fying

    nutrient requirements

    offish.

    J.

    Fish.

    Res. Board

    Can.

    33: 167-172.

    Four replicates

    of50

    rainbow

    trout(Salmo

    gairdneri)frngerlings each

    were

    assigned

    o each

    of

    seven

    dietary

    protein

    levels from 30 Io 60% in 57o

    ncrements. Percentage weight

    gains

    for

    10

    wk were related to dietary protein levels using an analysis ofvariance and multiple comparisons

    of

    the means,

    a broken line analysis, and a

    polynomial

    regression analysis. When

    correctly

    used, one

    ofthese techniques should

    lead to

    a

    reasonable conclusion concerning

    dietary

    protein

    requirements.

    However, the

    polynomial

    regression analysis has the

    advantage of being

    continu-

    ous, like

    the relation ofgrowth

    to

    dose, and should be

    more accurate than

    the other methods

    when

    the intervals

    between experimental

    dietary

    nutrient

    concentrations

    are

    wide.

    The

    polynomial

    regression

    analysis also

    provides

    the

    basis

    for

    making economic decisions

    relative to

    protein

    requirements

    or maximum

    economic

    returns.

    Zer rouN,I .

    H. , D. E. Ul ln rv,

    W.

    T. M,, rcep, ,

    .

    L. Grr- l , .qNo W. G.

    BencrN. 1976.

    Quanti fying

    nutrient

    requirements

    offish. J.

    Fish. Res. Board Can. 33: 167-1'72.

    De s

    groupes,

    contenant chacun 50

    alevins de la

    grosseur

    du doigt

    de truite arc-en-ciel

    S

    almo

    gairdneri),

    ont 6t6 expos6s dans

    quatre

    essais

    r6p6t6s ir chacun de sept

    niveaux de

    proteines

    alimentaires allant de 30 d 60%, par gradins de 5Va.Les gains pond6raux en pourcentage ont 6t6

    trait6s

    par

    analyse de

    variance

    et comparaisons multiples

    de

    moyennes,

    ainsi

    que par

    analyse

    de

    ligne

    bris6e

    et

    analyse

    de r6gression

    polynomiale.

    Lorsque utilis6e correctement,

    I'une

    ou

    I'autre

    de

    ces m6thodes devrait

    permettre

    d'en arriver

    d une

    conclusion

    raisonnable

    quant

    au x

    exigences

    en

    prot6ines

    alimentaires.

    Cependant, I'analyse de

    r6gression

    polynomiale

    a

    I'avantage

    d'Otre continue, tout comme

    la

    relation entre

    la

    croissance

    et

    la

    dose, et devrait

    Btre

    plus pr6cise

    que

    les autres

    m6thodes

    lorsque les extrmes de concentration des

    substances

    nutritives

    du

    r6gime

    exp6rimental sont espac6s. De

    plus,

    I'analyse de r6gression

    polynomiale

    peut

    servir

    de base d des

    d6cisions 6conomiques ayant

    trait aux

    prot6ines

    requises

    pour

    des

    profits

    maximums.

    Received

    une , 1975

    Accepted

    October2l, 1975

    EsrIlraerEs

    of

    quantitative

    nutrient requirements

    are influenced

    not

    only by the criteria used but by

    the

    statistical

    methods

    chosen

    to

    evaluate

    criterion

    response

    to differing

    dietary

    nutrient

    concentra-

    tions.

    In

    studies of

    protein

    or

    amino

    acid

    require-

    ments

    of fish,

    confusion is evident regarding

    the

    appropriate

    method

    to use

    (Delong

    et al. 1958,

    1962;

    Halver

    1960,

    1965;

    Halver

    et

    al. 1964;

    Chance

    et al. 1964;

    Cowey

    et aL.

    1972;

    Zeitow

    et al.

    1973. 1974).

    This matter

    is

    of

    considerable

    Regu

    e 4juin 1975

    Acceptde 2l octobre

    975

    significance due to the substantial

    growth

    of fish

    culture

    (Bardach

    etal.1972;

    Huet

    1973)

    and the

    observation

    that nutritional deficiency

    diseases

    are

    common

    wherever intensive flsh culture

    is

    prac-

    ticed

    (Halver

    1972).

    The

    qualitative

    nutrient

    re -

    quirements

    of

    various

    fish

    species

    have been

    previously

    reviewed

    (Halver

    1969,

    1970, 1972;

    Phil ips

    1969; Cowey

    and Sargent 1972). When

    estimating

    nutrient

    requirements

    to

    achieve

    bio-

    logical

    potential,

    the consequences

    of using

    various

    statistical

    techniques should be considered, and

    several of these techniques

    will be discussed.

    Weight gain is

    a common

    parameter

    used in

    evaluating different dietary levels of an essential

    nutrient

    and in estimating the

    quantitative

    require-

    ments

    of that nutrient

    for fish.

    While

    weight

    gain

    'Michigan

    Agricultural

    Experiment

    Article

    No. 6929.

    Printed n

    Canada

    I3861)

    Imprim6

    au Canada

    (J3861)

    Station

    Journal

    167

    Forpersonaluseonly.

  • 8/11/2019 Quantifying Nutrient Requirements of Fishl

    2/6

    r68

    J. FISH.

    RES. BOARD

    CAN.,

    VOL.

    33(I) , 1976

    Tanrr

    1. Surnmary

    of various responses

    f rainbow

    trout

    (Salmo gairdneri)

    fingerlings to different

    levels

    of dietary

    protein

    concentration

    derived

    rom

    Zeitoun

    et al 1973).

    Dietary

    protein

    concentration

    (%,)

    SEMg

    0

    5

    0

    s

    0

    5

    0

    Initial

    weight

    (s')

    Final

    weight

    g)

    Weight

    eain

    (:Z)^

    Weight

    gain (f)b

    Protein

    etention

    g)

    6 .5

    14.2.

    1 1 9 . 3

    119.3 '

    t . z -

    6 . 8 o

    7 7 O d

    1 5 0 . 7 d

    I

    5 0 .7d

    1 . 6 0

    6 . 2 .

    1 8 . 5

    197

    3{

    197 3

    2 . 0 .

    6 . 3 '

    19.2d

    204 s

    204.5

    2 7 f

    6 . 4 '

    79.6d

    208.9

    208.9'

    2 . 2 f

    6 . 3 ,

    7 2 d

    0 . 1 5

    t9 .4d

    20.5t

    0 .39

    2 0 8 . 1

    1 8 4 . l f

    7

    l

    2 0 8 . 1

    1 8 4 . 1

    7 . l

    2 . 2 d

    2 . 3 f

    0 . 0 6

    uDuncan's

    multiple

    range

    est

    was

    used

    o compare

    he means.

    bTukey's

    Honestly

    SignificantDifference

    HSD)

    test was

    used o compare

    he

    means.

    'd' 'rMeans

    n the

    same ine

    followed

    by different

    superscripts

    re significantly

    P

  • 8/11/2019 Quantifying Nutrient Requirements of Fishl

    3/6

    ZEITOUNET AL.:

    QUANTIFYING

    NUTRIENT

    REQUIREMENTS

    Tantn 2. Mean

    square errors

    (MSE)

    of different

    regression ines for

    percentage

    weight

    gain

    of rainbow trout

    fingerlings.

    Range of dietary

    protein

    levels

    ested

    f)

    Regression

    equation

    169

    MSE

    3V40.7

    30-40. .

    30-40.9^

    30-41

    0

    30-60b

    Y :

    - 1 2 0 . 9 + 7 . 9 1 X

    Y :

    - 1 1 8 . 6 + 7 . 8 4 X

    Y

    :

    -116 .3+7 . ' 7 ' 7X

    Y :

    - 1 1 4 . 0 + 7 . 7 0 X

    Y

    :

    -

    39796

    24.58X-0.2476X2

    310 .63

    310.47

    310 .46

    310.

    60

    278.20

    uFirst

    order

    polynomial.

    Valuesaboveeach

    breakpoint

    equated o

    the

    break-

    point

    value.

    Breakpoint

    valuesaboveand

    below hose isted

    were ested,

    nd

    as

    the

    breakpoint

    deviated

    urther from

    40.9,

    the

    mean

    square

    error increased.

    bSecond

    order

    polynomial.

    30

    35

    40

    45 50

    55 60

    olrfory

    Piotlin

    Concanfrolion,

    6

    Ftc. 1. Effect of Duncan's and Tukey's tests on the

    point

    of interception

    of the horizontal line and the

    percentage

    weight

    gain

    of rainbow trout

    (Salmo

    gairdneri)

    fingerlings.

    Yp is

    the average

    percentage

    when

    Duncan's

    test was used

    to

    compare the means

    and Yr is

    the average

    percentagegain

    when Tukey's

    test

    was

    applied.

    the

    two

    methods

    is different,

    and one

    can

    expect to

    obtain different

    results

    in some

    sets

    of

    data.

    BnorpN

    LrNe Axlrysrs

    Baker

    et al.

    (1.971)

    used

    the

    broken

    line

    analysis

    to estimate he requirements

    of tryptophan

    for

    baby

    pigs.

    Thiq

    procedure

    was developedby

    H. W.

    Norton,

    University

    of

    Illinois, Urbana, 1972,

    personal

    com-

    munication, and assumes

    a

    positive

    linear relatron

    between

    growth (Y)

    and the dietary level of the

    essentialnutrient (X) at or below the minimum re-

    quirement.

    At the

    minimum

    requirement

    level

    (R),

    the ascending

    ine

    which represents he relation be-

    tween

    growth

    and diet

    nutrient concentration

    breaks

    instantly

    to

    horizontal.

    Assuming that the

    breakpoint

    representsR, then Y

    values for X at and

    greater

    than

    R are

    estimates

    of

    the same response.

    A1l

    values

    of

    X

    greater

    than the breakpoint are

    equated

    o

    the

    breakpoint

    R for

    purposes

    of comput-

    inga regressionine,

    Y

    =

    a* bX.

    Several

    egressions

    are attempted using

    different breakpoints.

    Concur-

    rently, the deviation

    sum of

    squares or

    each

    regres-

    sion

    is estimated.The

    regression ine that

    gives

    the

    least mean

    square

    error

    (MSE)

    is considered he

    best

    fitted line

    to

    represent

    the linearity between dose

    levels and responsesand

    which is the least

    squares

    estimate

    of the

    requirement.

    Using the

    growth

    data of

    rainbow

    trout

    fingerlings

    from Table 1

    for

    this

    analysis,

    and

    using arbitrary

    values

    of X at O.lVo

    dietary

    protein

    intervals, the

    regression

    ine

    that

    fit the

    percentage

    weight

    gain

    with

    a

    breakpoint

    at

    40.9Vo

    protein

    in

    the

    diet

    gave

    the

    least MSE if compared to

    other

    lines

    (Table

    2) .

    Que.one,uc

    RtcnpssroN

    on Sr.coNo Onosn

    Porvvo-

    UIAI- RECNISSION ANALYSIS

    The second

    order

    polynomial

    regression

    analysis s

    represented y the equation Y : Bo * B.X * B X'.

    This

    curve is characterized

    by having

    a

    unique

    maxi-

    mum

    point

    (Y-*)

    along

    its

    range.

    The

    value

    of

    X- *

    that correspo nds o

    Y-.* is defined as the

    maximum

    concentration of the

    dietary nutrient that

    produces

    optimum

    growth,

    and

    beyond

    which

    growth

    is

    de-

    pressed.

    The

    advantage

    of

    this

    procedure

    is that it

    often

    provides

    a

    better empirical

    fit

    to

    the

    growth

    responses

    of

    living organisms which

    do not exhibit

    an abrupt change

    rom linearity

    as

    postulated

    n

    the

    previous

    wo analyses.

    Cowey et al.

    (t972)estimated

    the

    protein

    require-

    ments of

    the

    plaice

    (Pleuronectes latessa)

    by apply-

    ing the

    quadratic

    regressionanalysis.

    The minimum

    dietary protein level was defined as the level that pro-

    duced the highest

    point

    on

    the

    curve. To apply the

    Forpersonaluseonly.

  • 8/11/2019 Quantifying Nutrient Requirements of Fishl

    4/6

    170

    J.

    FISH,

    RES. BOARD

    CAN..

    VOL.

    33(I\ . 1976

    same

    concept

    o rainbow

    trout fingerling

    growth

    data

    (Table

    1),

    a

    quadratic

    equation

    was

    calculated

    (Y

    :

    -397.96

    +

    24.58X

    -

    0.2476X

    which

    showed

    that

    the maximum

    response

    of these fish

    was

    achieved

    at

    50Vo

    dietary

    protein (Fig.

    2).

    In

    addition, the

    profile

    of

    the data,

    which

    exhibited

    a substantial

    growth

    depression at both extremes of dietary protein con-

    centration,

    favored

    the use

    of this

    analysis.The cal-

    culated

    MSE

    of the

    curve

    was noticeably

    lower

    (278.20)

    than

    those

    estimated

    by the broken

    line

    procedure

    (Table

    2) .

    Although

    this

    curve would

    best represent

    he rela-

    tion

    of

    $owth

    to

    dose, t

    does not reflect

    the

    practi-

    cally insignificant

    differences

    in

    percentage

    gain

    below

    and beyond

    the maximum

    point,

    nor

    does it

    consider

    he

    ability

    of the animal

    to adapt

    to a range

    of dietary

    protein

    levels between

    a deficiency

    on the

    left

    side

    of the

    curve

    and toxicity

    on the right.

    That

    is,

    there

    are minimum

    and maximum

    levels

    of

    intake

    within

    which

    the

    animal is

    able to

    store, excrete,

    or

    adapt to the level of nutrient supplied without

    sub-

    stantial

    changes

    n metabolic

    processes.

    A

    statistical

    approach

    could be

    adapted

    o determine

    the level of

    nutrients

    in

    the diet that

    can

    yield

    a response

    hat is

    within

    a certain

    confidence

    range.

    of Y^ .

    The

    selection

    of the

    confidence level

    is

    dependent on

    the

    researcher

    and his

    objectives. n

    the case

    of the rain-

    bow

    trout fingerlings,

    confidence

    imits of

    95Vo of

    all estimated

    esponses

    f the curve

    (?)

    were

    calcu-

    lated

    using

    hc fol lowing

    expression:

    Y

    +

    t . . . L

    n

    + ( X - X ) ' V ( b , )

    +

    ( Z - Z ) ' V ( b )

    + 2(X - XltZ _ 71corlb,b,1f

    where

    MSE is

    the mean square

    error of the means,

    n is

    the number

    of observations

    on Y, X is the treat-

    ment level,

    X is

    the average

    of treatments, Z is

    the

    square

    of X,2

    is the

    average f Z,

    andV(b,), V(0,),

    llo

    l2o l3o t4o t5o

    160 r7o

    rao r9o

    2oo 2to 22o

    2fi

    W.ight

    coin, %

    Frc. 3. Relation

    between

    percentage

    weight

    gain

    and

    protein

    retention of rainbow

    trout

    fingerlings.

    and cov(brDr) are variances and

    covariance

    of esti-

    mated

    parameters

    of the

    polynomial

    curve.

    Levels

    of confidence

    were

    plotted (Fig.

    2) on either

    side of the

    quadratic

    line,

    and

    a straight line

    parallel

    to

    the

    abscissa and

    passing

    through the maximum

    level

    of the lower

    line of the confidence limit was

    drawn.

    Moving left from Y-.., this horizontal line

    first crosses he

    polynomial

    curve at a

    point

    Xr.

    Then,

    continuing

    to the left, the

    horizontal line intersects

    the

    upper

    line

    of

    the confidence limit at

    a

    point

    Xo.

    The value X' is the estimated level of dietary

    protein

    concentration expected to

    produce

    a

    growth

    response

    equal to the lower

    bound

    on the

    interval

    estimate of

    Y-

    in

    the

    initial

    study.

    The value Xo is the

    estimated

    level

    of

    protein

    concentration

    for which

    an upper

    bound on predicted growth response is equal to the

    lower bound on Y- *. Strictly speaking one cannot

    say that the responseat

    X

    (say

    Yr)

    is not

    statistically

    significantly different

    from

    Y^ ,

    because

    the dis-

    tribution of

    (Y* *

    -

    Y')

    is

    not

    known

    exactly and

    the sampling

    variance

    is

    a

    complex nonlinear func-

    tion. But

    statistical significance

    s

    somewhat

    irrelevant

    for

    regressionswith significant

    parameters (Williams

    1959).

    What

    is

    relevant

    s

    a difference hat

    has

    prac-

    tical importance.

    When economics dictates, dietary

    levels

    of

    protein

    ranging between

    Xo

    and

    Xr

    may be

    taken

    as a

    rough

    estimate

    of the concentration which

    minimizes cost while maintaining adequate

    growth

    response. f

    estimatesof

    cost

    per

    unit dose

    (C)

    and

    economic ncome per unit of yield (I) are available,

    one can

    base

    a decision on those rather than

    the

    somewhat arbitrary range,

    &

    to X'.

    Savings

    from

    reduced

    dose are C(X- -

    -

    X ),

    where the

    economic

    dose

    (X )

    is

    to

    be

    determined. Loss

    in

    return is

    IAY-, , where

    yield

    expressed at the

    economic dose

    is

    (1

    -

    A)Y---. If

    one maximizes net return

    with

    respect o A,

    then

    Xu

    =

    X- *

    -

    {A[X'9- *

    (bo/b )l]'/'

    and

    A

    -

    (c/2r),[x,^^*

    -

    (bo/

    b )]/Y,^^*.

    Letting

    C- *

    =

    CX- * and

    I*

    :

    IY- *,

    one obtains

    X

    =

    X* -[1

    -

    C^ /2I^^*)

    + c- /2I- -X ^,*) (b, /b ) l

    as the

    estimated economic dose.

    Whereas

    it is

    possible

    2. 4

    '6

    I A

    220

    zto

    200

    t20

    lo

    roo

    t90

    8 l @

    .E

    tzo

    _

    160

    I

    r5o

    t4 0

    t30

    30

    35 40

    xo xr

    45

    r@r

    55 60

    DialoryProll in

    Concantrol ion,. [

    Frc. 2.

    The

    second order

    polynomial

    relation

    (solid

    curved

    line)

    of

    percentage

    weight

    gain

    and dietary

    protein

    concentration

    for rainbow

    trout fingerlings,

    with

    0.95 confidence

    limits

    (dashed

    curved

    lines).

    O Estimated mean of Y (Y); Observedmean of

    Y.

    Forpersonaluseonly.

  • 8/11/2019 Quantifying Nutrient Requirements of Fishl

    5/6

    ZEITOUN

    ET

    AL.:

    QUANTIFYING

    NUTRIENT

    REQUIREMENTS

    t7 l

    to set fiducial limits

    on

    X* *

    (Finney

    1964),

    which

    is economically irrelevant,

    the

    complexity

    of

    X makes

    computation

    of limits intractable.

    Discussion

    An examination

    of

    the

    consequences

    of

    using

    the

    three described statistical

    procedures reveals

    that growth

    maxima may

    be

    identified

    in a

    range

    of dietary

    protein

    concentration

    from

    40 to 50%.

    llowever,

    selection of the upper

    figure

    assumes

    that

    the maximum

    percent

    weight

    gain deflned

    by application

    of the

    second order

    polynomial

    regression

    represents a

    response which is

    different

    biologically from the lower values

    defined by

    use

    of

    the

    polynomial

    limits.

    Statistically,

    the

    limits

    of

    response

    expected with

    sOEo

    (X-u*)

    dietary

    protein include the mean response expected with

    44Vo

    (Xr)

    dietary

    protein.

    Economically,

    this

    difference in dietary

    protein

    can be very

    impor-

    tant,

    particularly

    if the saving is

    greater

    than

    the

    fall

    in returns

    (Carpenter

    L97l).

    Dietary

    protein

    is

    generally

    one of the most

    expensive compo-

    nents

    of

    artificial

    fish

    diets, and if increasing

    increments

    of dietary

    protein

    concentration

    do

    not

    result in concomitant increase in value,

    the

    economic

    successof fish

    culture

    is

    in

    jeopardy.

    The

    selection

    of

    Xo

    to

    X1 on the

    polynomial

    regression

    as the

    minimum

    range

    of

    protein re-

    quirement

    is

    an

    economic decision; and

    while

    this

    decision

    may

    be economically conservative,

    the

    selection of X-o- is more

    physiologically

    con-

    servative.

    Using local

    prices

    of

    dietary

    protein

    and

    fish, average initial weight of the

    fish,

    and

    estimated

    efficiency of

    feed

    conversion,

    it was

    possible

    to

    estimate the ratio, C-o,/2I,,,u ,

    which

    equals

    0.

    1534.2

    Then,

    the

    estimated

    economic

    protein

    requirement is

    X

    -

    50(1- 0.L534)

    +

    (0 .1534 /

    50)

    ( -397.96/

    -0 .25)

    =

    47 .2%.

    For

    the

    current data,

    this

    result

    is

    above Xt

    -

    44Vo.

    However, if

    the cost

    of dietary

    protein rises

    above the figure

    assumed in these

    calculations, or

    if the selling price of fish declines, then the esti-

    mate

    of

    the economic

    protein

    requirements

    (o r

    economic

    dose, X,) would

    be

    lower. Decreased

    efficiency

    of conversion of dietary

    protein

    to

    gain

    'Protein

    was

    derived

    from Purina Trout Chow

    (40%

    minimum

    crude

    protein),

    at

    a

    price

    of

    $9.85/

    22.7

    kg

    or

    $1.08/kg

    of

    protein.

    Local

    prices

    of

    farm-reared

    rainbow

    trout averaged approximately

    $2.64/kg.

    Estimated

    efiiciency of

    feed

    conversion

    to

    gain

    was 1.5. Cost

    of

    protein

    per

    kilogram

    gain

    at

    the

    m a x i m u m

    ( X - , *

    :

    5 O % )

    :

    ( 1 . 5 ) ( 0 . 5 ) ( $ 1 . 0 8 )

    0.81;

    gain

    at

    maximum

    rate

    (average

    nitial

    weight)

    lY^- /100)

    -

    l l

    =

    (0 .2)1212/100) 1 l

    =

    0.224ke;

    maximum

    cost

    (C, , )

    -

    (0.81)(0.224)

    =

    0.1,874;

    I- -

    -

    (2.64) 0.224)

    :

    0.5914.

    will have the

    same effect.

    Estimation

    by Xn is to

    be

    preferred, but when

    economic considerations

    are strong,

    and

    information

    on costs

    and income

    are

    difficult

    to obtain,

    one may

    choose

    as a

    simple

    substitute

    the

    range between

    Xe

    and

    X1.

    Certainly an increase in the number of observa-

    tions used

    to

    determine

    the

    polynomial curve

    would

    narrow

    the confidence

    limits

    and tend to

    shift the range

    of Xn

    to Xr

    to the right. Whether

    this

    shift is important

    must be

    answered

    by

    the

    researcher

    himself.

    If the

    criterion

    of response

    were badly chosen

    (in

    representing

    the effect

    of

    increasing

    dietary nutrient

    levels),

    such a shift

    may be very

    important.

    Weight

    gain has

    been

    criticized

    (Phillips

    et al.

    1957; Allison

    1959; Maynard

    and

    Loosli 1969)

    for inaccuracy

    as

    a

    measure of

    growth,

    since

    gain

    in

    weight

    may

    result from

    deposition

    of fat rather

    than

    from true

    growth.

    The

    potential for

    growth

    may be

    considered

    identical

    with

    maximum

    pro-

    tein

    retention,

    and

    when

    values for the latter

    (Table

    1)

    were

    regressed

    against

    percent

    weight

    gain (Fig.

    3

    ),

    the correlation

    coelncient

    was

    0.93. It is

    apparent

    that

    for rainbow

    trout finger-

    Iings, weight

    gain

    was

    a

    good

    measure

    of

    true

    growth. Analysis of

    variance

    and

    multiple com-

    parison

    of

    treatment

    means,

    broken line anal-

    ysis,

    and

    second

    order

    polynomial

    regression

    analysis may

    lead to

    similar conclusions

    concern-

    ing

    dietary

    protein

    requirements.

    Flowever, the

    polynomial approach has the advantage of being

    continuous,

    like the

    relation

    of

    growth

    to dose,

    and

    should

    be more

    accurate than

    the other

    methods if the intervals

    between

    experimental

    dietary

    nutrient

    concentrations

    are

    wide. Also,

    the

    polynomial

    method

    is well

    adapted to

    eco-

    nomic analysis

    if

    information

    is available

    on

    costs and

    returns. One

    should

    remember,

    how-

    ever, that the

    polynomial is

    only

    a smooth, sym-

    metric approximation

    to the real

    relation of

    gain

    and intake and may

    be

    in some cases

    better than

    others.

    Acknowledgments

    We thank

    Drs P.

    L Tack

    and

    E.

    R.

    Miller

    fo r

    reviewing he

    manuscript.

    ArrrsoN,

    J.

    B. 1959. heefficiency

    f utilization f

    dietary

    proteins, .97-116.

    n A. A. Albanese

    ed.]

    Protein

    and

    amino

    acidnutrition.

    AcademicPress, nc.,

    New

    York. N.Y.

    BexE.n,D. H.,

    N.

    K.

    AII-rN, J.

    Boovceenor, G.

    GrqBsn,

    .qNnH. W.

    NoxroN. 1971.

    Quantitative

    s-

    pects

    of

    D- andL-tryptophan

    utilization

    by

    the

    young

    pig.

    J.

    Anim.

    Sci.33: 246.

    Bnnrncn, J.

    E.,

    J.

    H.

    Rvruen, eNo W.

    O. Mcl-enNev.

    1972.Aquaculture: he farming and husbandry of

    freshwater

    and marine organisms.

    Wiley-Inter-

    science, ew York,

    N.Y.

    868

    .

    Forpersonaluseonly.

  • 8/11/2019 Quantifying Nutrient Requirements of Fishl

    6/6

    172

    J.

    FISH. RES,

    BOARD

    CAN,, VOL. 33( I ) , 1976

    Ce.reeNren,

    K.

    J. 1971.Problems

    n

    formulatingsimple

    recommended

    llowances

    f amino

    acids

    or

    animals

    and

    man.Br.

    J. Nutr.

    30:

    73-83.

    CneNcE,

    R.

    8.,

    E. T.

    Mrnrz.,cNo

    J. E.

    Helvrn.

    1964.

    Nutrition

    of

    salmonoid

    fishes:

    XII.

    Isoleucine,

    leucine,

    valine

    and

    phenylalanine

    equirements

    f

    chinooksalmon nd nterrelations etweensoleucine

    and

    eucineor

    growth.

    . Nutr .

    83:

    177-185.

    Cowrv,

    C.

    B. ,

    J . A. Popp,

    . W. AonoN, lNo

    A. Br_ l rn .

    1972.

    Studieson the

    nutrition

    of

    marine

    latfish.

    The

    protein

    equirement

    f

    plaice Pl

    eu

    one

    es

    plat

    es sa).

    Br.

    J. Nutr. 28:447456.

    Cowey,

    C.

    B.,

    elro

    J. R.

    SencrNr.

    1972. ishnutrit ion.

    Adv.

    Mar. Biol.

    0:

    383492.

    DeLoNc,

    D.

    C., J. E. Huvpn,

    AND

    E. T.

    Mrnrz.

    1958.

    Nutrition

    of

    salmonoid

    ishes:

    VI.

    Protein

    require-

    ments

    of chinook

    salmon

    at two

    water emperatures.

    J. Nutr.

    65:589-599.

    1962.Nutrition

    of

    salmonoid

    ishes:X.

    Quantita-

    tive

    threonine

    equirements

    f

    chinooksalmon t

    two

    water emperatures.. Nutr. 76: 17.4-178.

    FrNNry,

    D. J.

    1964.

    tatist ical

    ethod

    n biological ssay.

    HafnerPublishing

    o.,New

    York, N.Y. 668

    .

    FtsnEn,

    H., D.

    JoHNsoN

    R.,AND

    G. A. LEvErr- 1e,.

    957.

    The phenylalanine

    nd

    tyrosine

    requirement

    f the

    growing

    chick

    with

    special eference

    o

    the

    utilization

    of

    the D-isomer

    of

    phenylalanine.

    J. Nutr.

    62:

    349-355.

    GnrurNcen,

    P.,

    H. M.

    Scorr, nNo

    R.

    M.

    FoRBEs.

    955.

    The

    effectofprotein

    level

    on the

    tryptophan equire-

    ment

    fthe

    growing

    hick.

    .

    Nutr.

    59:67-76.

    Helven,

    J.

    E.

    1957.

    Nutrit ion

    of

    salmonoidishes:

    IL

    Water-soluble itamin

    requirements

    f

    chinook

    sal-

    mon.

    J. Nutr.

    62:225-243.

    1960.Vitamin and amino acid requirements f

    salmon.

    roc.

    Fifth nt.

    Congr.Nutr.

    191:

    2.

    Abstr.)

    1965.

    Tryptophan

    requirements

    of chinook,

    sockeye.and

    ilver

    almon.

    ed.

    proc.

    Fed.Am.

    Soc.

    Exp.

    Biol.24:229.

    Abstr.)

    1969.

    itamin

    equirements,

    209-232.ln

    O.W.

    Neuhaus

    nd

    J. E. Halver

    [ed.]

    Fish in r esearcn.

    Academic

    ress,

    nc. ,

    New

    york,

    N.y.

    1970. utr i t ion

    n

    marine

    quiculture,

    .75-102.

    1n

    W.

    J. McNeil

    [ed.]

    Marine

    aquiculture.

    Oregon

    State

    UniversityPress,

    Corvallis,

    Oreg.

    1972.

    The vitamins,

    p.

    29-103.1nJ. E. Halver

    [ed.]

    Fish

    nutrition.

    Academic

    Press, nc.,

    New York,

    N.Y.

    Her-vpn, J. E., L.

    S. Bares,

    AND E. T. Meptrz.

    1964.

    Protein equirements

    or

    sockeye almon nd ainbow

    trout.

    Fed.

    Proc.,Fed. Am.

    Soc.Exp. Biol.

    23: 17'18.

    (Abstr.)

    Huer,

    M. 1973.Textbook

    of

    fish culture. Fishing

    News

    Books,

    td. ,

    London,

    36

    .

    LAGLER,

    .

    F., J. E. Bnnoecu,

    euo

    R. R.

    Mrr-lpn.

    1962.

    Ichthyology.

    ohn Wiley

    and Sons,

    nc.,

    New

    York,

    N .Y . 545

    .

    MnvNeno,

    L. A., e.No

    . K. Looslr.

    1969.

    Animal

    nutri-

    t ion.

    McGraw-Hil l ook

    Co.,New York, N.Y.

    613

    .

    Mlrcnrr-r,

    H. H.1962.

    Comparative

    utritionof man

    and

    domestic nimals.

    Vol.

    l. AcademicPress, nc.,

    New

    York,N.Y.701

    .

    MonnrsoN,

    W. D., T.

    S.

    Hl.vrt-roN,

    ANDH. M.

    Scorr.

    1956.Utilization of D-tryptophan

    by the chick.

    J.

    Nutr. 60:47-63.

    PnrlLIrs,

    A.

    M.

    Jn. 1969.Nutrition,

    digestion ndenergy,

    p.391-432.1n

    W. S.

    Hoar

    andD.

    J.

    Randall

    ed.]

    Fish

    physiology.

    ol. 1.Academic

    ress,

    nc. ,

    New York,

    N . Y .

    PHILltrs,

    A. M. Jn.,D.

    R. BnocxwAy,

    F. E. Lovpr-ncn,

    AND

    H.

    A. Pooolrer.

    1957.A

    chemical

    omparison

    of hatcheryand

    wild brook

    rout. Prog.Fish

    Cult. 19:

    19-25.

    SNeoecon,G. W. 1956.Statisticalmethods. owa

    State

    University

    ress, me s, owa.

    534

    .

    Srn.pr-, . G. D., eNo

    J.

    H.

    Tonnlp. 1960. r inciples

    nd

    procedures

    f statistics.

    McGraw-HillBookCo.,

    New

    York, N.Y.

    481

    .

    Wllltarls, E. J. 1959.Regression nalysis. ohn Wiley

    and

    Sons,

    nc . ,

    NewYork,

    N.Y.214p.

    ZrrrouN,

    L

    H.,

    J. E. Her-vrn,

    D. E.

    ULLREy,

    ND

    P. L

    Tecr. 1973. nfluence

    of salinity

    on

    protein

    equire-

    mentsof

    rainbow

    rotl(Salmo gairdneri)

    ingerlings.

    J. Fish.

    Res.

    BoardCan.

    30:1867-1873.

    ZerrouN,

    L

    H., D. E.

    ULLREv,

    .

    E. Halvrn, P.

    .

    Tecr,

    AND

    W.

    T.

    MecEe. 1974. nfluenceof salinity on

    protein

    equirements

    f coho salmon

    Oncorhynchus

    k isutch)smol ts .

    J .

    F ish.

    Res. Board Can.31:

    I 145-1 48.

    Forpersonaluseonly.