Lecture 3

Age structured models:

Recruitment, growth, and mortality
(exponential decay) models
 Age structured models
• Note that SPM, developed from the logistic population growth model,
treat a stock as undifferentiated biomass and ignore sexual, size and
age based differences.
• Age structured models account for the processes of
o recruitment,
o growth
o mortality.
• In age structured models, the population can be subdivided into
cohorts so that the dynamics of each cohort are followed separately,
• In this and the other following lecturers, the age structured models
will be covered
The stock recruitment relationship
• Recruitment refers to life-stage that first becomes vulnerable to fishing
gear
• For purposes of stock-recruitment relationship recruitment is defined as
the population still alive at any given time after the egg stage.
• Recruitment requires some adult fish in the fishery and it depends on
fecundity and abiotic factors
• Recruitment overfishing
o shows stock is fished so hard that its size is reduced below a level that can
produce enough new recruits to replace those dying.
o Detected through the relationship between spawning stock size and recruitment
levels
 The stock recruitment relationship - contd

• Spawning stock biomass refers to a measure of reproductive

productivity of the mature population.
• Reproductive capability can be estimated in descending order of
reliability from:
o Average fecundity by age and the proportion of each age
o The number of mature females multiplied by the average fecundity
o Total biomass of mature individuals
o An index of abundance for the population in the year of spawning
 Recruitment Models
• Several models have been used to suggest the relationship between
recruitment and stock size
• Examples of mathematical models of the stock – recruitment
relationships include:
• Beverton – Holt recruitment Model
• Ricker Model
• Shepherd recruitment model
• Deriso and Schnute Model
Beverton – Holt recruitment Model
• Incorporates density dependent survival rates reflecting cohort competition
for critical resources

R = {S/(α+βS)}eЄ
Where:
• R: recruitment
• S: the measure of spawning stock size
• α and β: parameters of the Beverton – Holt relationship. The β value determines the
asymptotic limit (1/β) while the differing values of α are inversely related to the
rapidity with which each value attains the asymptote
• eЄ : indicates that residual errors between the relationship and observed data are
expected to be lognormal.
Beverton – Holt recruitment Model
• Incorporates density dependent survival rates reflecting cohort
competition for critical resources
• There are a few formulations used for the Beverton-Holt stock
recruitment relationship but the commonly seen is:
R = {aS/(b+S)}eЄ
• Where:
• a is the maximum number of recruits produced (the asymptote)
• b is spawning stock needed to produce, on average, recruitment equal to half
maximum
Beverton – Holt recruitment Model - contd

A comparison of
Beverton-holt
curves relating
recruitment against
spawning stock size
using equation: R =
{aS/(b+S)}eЄ with
different values of
b. In all cases the
value of a = 10000
which is the
maximum
recruitment
possible. The
straight dotted line
is the line of
replacement.
Ricker Recruitment Model
• There are different formulations of the Ricker Equation, but the
commonly used is:
R = aSe–bseЄ
• Where
o R = recruitment
o S = the spawning stock
o a = the recruits-per-spawner at low stock levels
o b = parameter that relates to the rate of decrease of recruits-per-spawner as S
increases
o eЄ is an indication that the residual errors between the relationship and
observed data are expected to be lognormal
 Ricker Recruitment Model - contd
Three Ricker
stock-
recruitment
curves based
on equation: R
= aSe–bseЄ.
Each of the
data series has
an a value of
10, with the b
value
indicated
Ricker Recruitment Model - contd
• Does not attain an asymptote but exhibits declining recruitment at
higher stock levels.
• Assumes that the density-dependent mortality term for the eggs and
juvenile stages relate to total stock size and the mortality term is
attributed to:
o Cannibalism of the juveniles by the adults
o Density dependence transmission of diseases
o Damage by spawning adults of each other’s spawning sites
o Density dependency growth combined with size dependent predation
 Shepherd’s recruitment model

• R = aS/[1 + (S/K)b]
• Where
o a = initial slope at the origin of the curve (maximum recruitment per unit
stock), which is attainable at low stock size
o S = spawning stock size
o K = the threshold stock size above which density-dependent effects dominate
density-independent effects
o b = compensation parameter (degree to which density independent effects
compensate for changes in stock size)
Shepherd’s recruitment model - contd

Shepherd curves for b = 0.5, 1, and 2, with a = 3 and K = 600.

Common attributes of the stock recruitment
models
• No real convincing models to handle the problem of recruitment.
• Large estimation errors often make recruitment appear to be
independent of spawning stocks
• There is need for more biological knowledge on the causes of natural
mortality in the recruitment models
 Growth Models
• Growth of individual fish in a stock leads to changes (increase) of fish
biomass.
• Methods used for fitting growth models are:
o age vs length,
o age vs weight,
o length vs weight.
• Many methods are used for ageing fish.
 Growth in length
• The most commonly used model is the Von Bertlanffy Growth Model that
describes growth in length as:
Lt = L∞{1-e-K[t-t0)}+є
• Where Lt = length at age t
o L∞ is asymptotic average maximum body length
o K is growth rate coefficient that determines how quickly the maximum is attained
o t0 is hypothetical age at which the species has zero length
o є denotes that residuals would be distributed normally about the expected growth line

Note: Please review the methods used for estimation of the parameters of the
VBGM
 Growth in weight
• Relationship between length and weight is described by a power function:

Wt = aLtb

• Where
• Wt =weight
• Lt = length
• b = allometric growth parameter (usually close to 3)
• a = scaling constant
• Von Bertalanffy equation for body weight can be derived from equations, Lt =
L∞{1-e-K[t-t0)}+є, and Wt = aLtb, to give equation:

Wt = W∞[1-e-K[t-t0)]b
Fish age determination

• Age of fish can be determined using the following methods:

o Use of growth zones on calcified (hard) parts of fishes: Involves interpretation
and counting of growth zones on calcified tissues such as: scales, otoliths, bones
e.g. opercula, vertebrae, fin spines.
o Use of length frequency distribution (LFD) also known as the Petersen method: .
Individual lengths of a large number of fish are taken
number of fish by length is plotted in LFD.
Age classes separate as peaks in the LFD.
Assumes the following: fish spawn annually with a short spawning season, the progeny grow
at a uniform rate, minimum overlaps in sizes of fish in adjacent age groups, distribution of
length within an age group is normal
o Tagging of fish (mark recapture):
o Fish are captured tagged and released in the environmental where the time of release and
re-capture is recorded accompanied by measuring growth rate parameters
o Observation of growth of fish raised in a controlled environment
o These fish are sampled periodically and measured to estimate size at age
Growth models:
Fish age determination - contd Use of Petersen Method

Use of calcified parts

Otoliths

Use of scales
Growth models
Estimation of the VBGM parameters
• The parameters of the VBGM (L∞, K, and t0) can be estimated when the following
information is known
o Age and length data combined
o Length measurements only
o Two or more length measurements obtained at different time periods
• The required information can be obtained from either of the following sources
o Data from resource surveys with a research vessel
o Data from samples taken from commercial catches
o Mark – recapture (tagging) experiments
• The following methods are used for estimating the parameters of the VBGM
o Fold Walford Plot
o The Gulland and Holt plot
o Chapman’s Method
o The von Bertalanffy Plot
Estimation of the VBGM parameters

Fold Walford Plot

• Used for quick estimates of
L∞ without calculations.
• L(t+dt) = a + b*L(t)
o where a = L∞ *(1-b) and
b=exp(-k*dt)
• The growth parameters K
and L∞ are derived from:
o K=-1/dt*lnb and
o L∞= a/(1-b)
Estimation of the VBGM parameters - contd

The Gulland and Holt plot

• Described by the equation:

dL/dt = a + b *
AvgL(t)
o where AvgL (t) = [L(t+dt) +
L(t)]/2
o The growth parameters K
and L∞ are obtained from K
= - b and L∞ = -a/b
o The Gulland and Holt Plot is
reasonable for small values
of dt
Estimation of the VBGM parameters - contd

Chapman’s Method
• Also assumes constant dt.
L(t+dt) – L(t) = C* L∞ – C*L(t)
which is linear and of the form
Y = a+bx
where
o a = C*L∞, b = -C and x = L(t)
o C = 1 – exp (-k*dt)
o K = -(1/dt)*Ln (1+b)
o L∞ = -a/b or a/c
von Bertalanffy Plot

• Can be used for estimating K and

t0 from age and length data.
o requires an estimate of L∞ as input
data
• The VBGM can be rewritten as
o -Ln (1-L(t)/L∞) = - K t0 + Kt
• linear relationship where the slope
b = K and the intercept a = - K* t0.
• more robust method than the
Gulland and Holt Plot (and the
Ford and Walford Plot) .
Tutorial 3: Estimation of the parameters of the VBGM for fish growth

1) Assume you have the theoretical age length

relationship of the Rhamphochromis esox (Boulenger, Age (years) TL (cm)
1908) from Lake Malawi presented in Table 1.
a) Using the information given in the Table, estimate the 1 11
growth parameters K and L∞ for the Von Bertalanffy
Growth Model using:
2 21
i. The Gulland and Holt Plot
ii. The Ford-Walford Plot 3 29
iii. Compare the values of the parameters obtained
from the Gulland and Holt Plot and the Ford-
Walford Plot.
4 34
b) Supposing that besides the information in Table 1, you
are given that L∞ for R. esox is 43 cm; 5 38
iv. Estimate the parameters K and t0 using the Von
Bertalanffy Plot. 6 40
v. Plot the Model using the parameters obtained from
the Von Bertlanffy Plot. (Use 1 year intervals on the 7 41
X axis for the age ranges of 0 to 10 years).
 Mortality
• Survival of individuals in marine fish population is affected by many
factors e.g.
o adverse conditions – e.g. storms and current preventing larvae reaching
nursery areas in marine fish
o lack of food – especially when pelagic larvae have exhausted their yolk-sac
reserves, and become dependent on surrounding food
o competition
o predation
• Mortality can best be understood by following up the numbers of fish
surviving in a cohort with age.
• Mortality of a cohort comprises of fishing and natural mortality.
 Exponential Decay model
• Loss of individuals in a population through death can be discussed in terms of
o the percentage of individuals that survive (survival rate) over a particular time interval,
or
o or the percentage that die (mortality rate).
• Number of individuals in the population at any time is given by
Nt = N0e-Zt
Where Nt = the number of individuals in the population after t years
N0 = initial population number
Z = instantaneous rate of total mortality
• The fraction surviving after one year is called the survival rate, S:
S = Nt+1/Nt = e-Z
• As percentage:
Survival (%) = 100 e−Z
Mortality (%) = 100(1 – e−Z)
 Exponential Decay model

• Note that the assumption is that after recruitment, adult mortality

rates are constant over the remainder of the life cycle,
o Though normally mortality rates decrease with increasing size
• In an exploited fish stock mortality caused by fishing (F) can be
distinguished from mortality caused by natural phenomena (M).
• Total mortality rate (Z) is the sum of instantaneous rate of fishing
mortality (F) and the instantaneous rate of natural mortality (M) i.e
Z = F+M and
Nt = N0e-(M+F)
Figure: Survival of fish subjected to different rates of fishing mortality
 Estimation of fish mortality rates
A number of methods are used for estimation of mortality rates e.g.
o Age based catch curve analysis
o Length based catch curves
o Use of mark-recapture data
o Use of data from unexploited fish stock surveys
o Use of environmental factors as well as evolved life history patterns
o Consideration of natural mortality rate required to reduce the numbers of fish
in a cohort to close to zero over the period of its life span
• We shall look at the age based catch curve analysis and use of
environmental factors. The other methods for mortality estimation
are well summarized by Spare & Venema (1998)
 Age based catch curve analysis

• Equation Nt = N0e-Zt can be log transformed to get a linear catch curve

as:
Ln(Nt) = Ln (N0) – Zt
• This Equation is a linear relationship known as catch curve whose
slope (absolute value) is the instantaneous mortality rate Z.
• A catch curve can be constructed by determining the age of individuals
in a single large sample of fish.
Assumptions of the catch curve analysis
• Age composition of the sample represents the age composition of the
stock. i.e. sample age composition regarded as “pseudo-cohort”
• All age groups are equally vulnerable to the fishing gear used for
sampling
• Recruitment and total mortality are constant
• The number of individuals in each age class decreases according to
constant mortality rate
Observations and recommendations regarding
age based catch curve analysis
• The assumption of constant recruitment can rarely be fulfilled in real life.
However, a cohort may be followed with time.
• Age composition of a single sample is justifiably used for constructing a
catch curve if the species in question has a long life span in which
recruitment varies in random fashion
• Use of catch curve to estimate instantaneous mortality is applicable to age
groups that are fully recruited
• CPUE can be used instead of numbers of survivors, as long as time series
data of CPUE and details of age composition are available.
• Age composition can be inferred directly from length-frequency data where
different age classes can be easily separated from each other.
 Use of environmental factors as well as evolved
life history patterns - continued
Pauly’s empirical formula
• Pauly, (1980) made a regression analysis of M (per year) on K (per year), L∞
(cm) and T (annual average temperature at the surface in 0C) based on data
from 175 different fish stocks and estimated the empirical linear relationship:

Ln M = -0.0152 - 0.279 * ln L∞ + 0.6543*ln K + 0.463*ln T

• Where
o M = natural mortality rate
o K = VBGM growth parameter
o L∞ = VBGM asymptotic average maximum body length
o T = surface water temperature
Tutorial 3: Estimation of fish mortality using
catch curve analysis
Age (years) Number caught
2) Using the theoretical
0 60
information in the Table for one
1 83
species of fish, use the catch curve
analysis to estimate the 2 164
a) total mortality rate of the fish 3 142
species. 4 87
b) Discuss the assumptions made 5 37
to estimate the total mortality 6 20
rate of the species.
7 12
8 9