Introduction to Fisheries Economics
by the use of Mathematica programming
Arne Eide Copyright c 2017 Arne Eide
Published by Publisher book-website.com
Licensed under the Creative Commons Attribution-NonCommercial 3.0 Unported License (the “License”). You may not use this file except in compliance with the License. You may obtain a copy of the License at http://creativecommons.org/licenses/by-nc/3.0. Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an “as is” basis, without warranties or conditions of any kind, either express or implied. See the License for the specific language governing permissions and limitations under the License.
First printing, March 2017 Contents
1 Introduction ...... 7
I Harvesting technology and population dynamics
2 Fishing effort ...... 11 2.1 Requirements of fishing 11 2.2 How to measure fishing effort? 11 2.3 The production of fishing effort 12 2.4 Introduction to elasticities 13 2.5 The Cobb-Douglas function 14 2.6 The CES function 15 2.7 Elasticity of scale 16
3 Fish catch production ...... 19 3.1 Catch production 19 3.2 Catch as a method of stock assessment 19 3.3 Cobb-Douglas catch production 23 3.4 Properties of fishing gears 23 3.5 Exercises 23
4 Population dynamics ...... 25 4.1 Basic principles 25 4.2 Classical surplus growth models 25 4.3 Depensatory growth 31 4.4 Age structured model 34 4.5 Exercises 39
5 The concept of equilibrium harvest ...... 41 5.1 Equilibrium harvest 41 5.2 Surplus production models 42 5.3 Age structured models 46 5.3.1 Numbered List...... 47 5.4 Remarks 47 5.5 Corollaries 47 5.6 Propositions 47 5.6.1 Several equations...... 47 5.6.2 Single Line...... 47 5.7 Examples 47 5.7.1 Equation and Text...... 47 5.7.2 Paragraph of Text...... 48 5.8 Exercises 48 5.9 Problems 48 5.10 Vocabulary 48
II Fisheries economics
6 The economics of catch production ...... 51 6.1 The economics of effort production 51 6.2 Exercises 54
7 Economic growth ...... 55 7.1 Labour creates capital 55 7.2 Neoclassical Economic Growth Theory 55 7.3 Exercises 56
8 Economic growth ...... 61 8.1 Labour creates capital 61 8.2 Table 61 8.3 Figure 61
Bibliography ...... 63 Articles 63 Index ...... 65
1. Introduction
Fishing, hunting and gathering food represent the oldest human forms of life, where people depended on wild food for subsistence. While the importance of hunting and gathering wild food decreased after the introduction of agriculture more than ten thousand years ago, fishing has over the last hundred years been increasingly more important as a source of food and wealth for millions of people around the world.
Fishing is an economic activity to produce food for own consumption, but today also for commercial markets and for the purpose of recreation and pastime. These activities often coexist but they could also be regarded as stages in economic and cultural developments, starting with the subsistence fishery, before developing a commercial industry and recre- ational use of the natural resource.
The aim of the earliest stage, the subsistence fishing, was to feed own household. Early barter economy allowed fish products to be traded but as a perishable product there were limits for the trade-ability of fish. Methods to preserve fish by drying, salting and smoking it, which contributed in developing fish markets where stored fish could be sold, were discovered and developed. Infrastructure then become the major obstacle, transporting the fish to the markets in the large cities demanded storing capacity, roads, vehicles for transport, but also organisations, agreements and security for those investing in the trade.
Increasing trade resulted in wealth creations, first in the business of trade and later also increasingly among fishers. A consequence of economic growth is that labour becomes relatively more expensive and capital more available, hence labour is substituted by capital as a consequence of economic growth, also in the fishing industry.
The aim of recreational fishing, the most recent utilisation of fish stock resources, is not primarily to produce catch, although the catch also often is utilised. The ultimate goals are however the adventure, utilising and enjoying the natural environment, sensing 8 Chapter 1. Introduction peaceful nature and experiencing relaxing activities. Recreational fishing has developed to become a major industry both in terms of accommodating the fishers and production of fishing tackle and equipment.
Fishing takes different forms in different societal and economic contexts. This book aims to discuss the economic activity of fishing through formal models, also including the dynamics of the utilised fish stocks. The analysis are based on basic microeconomic principles and standard theories of fisheries economics are presented.
The book includes however also some attempts of employing methodology which are more rarely seen in other bioeconomic textbooks. The book offers an introduction to programming in the Wolfram Language, presenting a number of relevant examples of how to use the software Mathematica as a tool in fisheries economics. Harvesting technology I and population dynamics
2 Fishing effort ...... 11 2.1 Requirements of fishing 2.2 How to measure fishing effort? 2.3 The production of fishing effort 2.4 Introduction to elasticities 2.5 The Cobb-Douglas function 2.6 The CES function 2.7 Elasticity of scale
3 Fish catch production ...... 19 3.1 Catch production 3.2 Catch as a method of stock assessment 3.3 Cobb-Douglas catch production 3.4 Properties of fishing gears 3.5 Exercises
4 Population dynamics ...... 25 4.1 Basic principles 4.2 Classical surplus growth models 4.3 Depensatory growth 4.4 Age structured model 4.5 Exercises
5 The concept of equilibrium harvest .... 41 5.1 Equilibrium harvest 5.2 Surplus production models 5.3 Age structured models 5.4 Remarks 5.5 Corollaries 5.6 Propositions 5.7 Examples 5.8 Exercises 5.9 Problems 5.10 Vocabulary
2. Fishing effort
2.1 Requirements of fishing Fishing is utilisation of a natural resource, the fish stock. In order to capture fish and other seafood products, proper catching methods have to be used. Some resources (shellfish, crabs, clams, etc.) could easily be picked on the beach or in shallow water when available, while other resources demand more elaborated catching techniques.
During the thousands of years humans have been utilising fish stocks for food, a large number of capturing methods have been developed. Some of them have proved to be efficient and are still in use, while others have been left behind. Today a large number of different catching technologies are in daily use, some of them known from ancient time while others only have been around for a few decades.
Angling is among the oldest fishing methods we know and is still used in commercial fisheries around the world. Hand line and long line fishing are the most important angling methods. Different types of fishing nets also have been in use since ancient times. Nets are used in modern fisheries as gill nets and in more sophisticated net constructions, as purse seines, Danish seines and trawling gears.
2.2 How to measure fishing effort? Fishing effort could be regarded as a commodity that could be produced in many different ways and in different quantities. As discussed above, it exists a vast number of different methods of catching fish. Their relative efficiency depends on a number of factors which we will discuss elsewhere in the book.
It is not obvious how to measure fishing effort, units that appear to be useful in some fisheries may be useless in others. In a fishery where the fishing fleet is fairly homogeneous, fishing effort may be measured in terms of 12 Chapter 2. Fishing effort
Gill net Purse seine
Trawl
Figure 2.1: Some standard fishing gears used in modern fisheries (Source: Wikimedia commons)
• Number of vessels per year • Number of fishing hours, days or trips • Towing hours (for trawl fisheries)
In the more common cases of heterogeneous fleets, useful fishing effort measures may be • Total numbers of engine horse power units • Total number of hooks (for hand line and long line gears) • Total number of nets (for gill net fleets) • Sum of vessel lengths • Sum of vessel tonnage
Each case must be evaluated on its own based on its own properties. The measurement method has high quality if all units are measured by the same scale, also over time. If performance data is available the best measure may be found by standardising an ef- fort unit based on combinations of several of the measures shown in the bullet points above.
Summary 2.2.1 — Units to measure fishing effort. There is no standard unit for the measurement of fishing effort.
2.3 The production of fishing effort In general all fishing effort (E) is produced in production processes involving labour (L) and capital (K),
E(L,K) = f(L,K) (2.1)
where labour and capital may substitute each other, producing the same amount of effort. That means that the same fishing effort may be produced by a large quantity of labour and a small quantity of capital as by a small quantity of labour and a large quantity of capital. 2.4 Introduction to elasticities 13
The variables Labour (L) and capital (K) in Equation 2.1 are referred to as produc- tion input factors, while the fishing effort (E) is the output from the production process.
Figure 2.2 shows three possible shapes Equation 2.1 may have. Each of the three cases gives a constant production of fishing effort for different combinations of labour and capital. The red broken line represents a situation where labour and capital not are substitutable factors, increasing one factor does not contribute in increased fishing effort if not the other factor also is increased. Capital ( K )
Labour(L)
Figure 2.2: Three production processes with different substitution elasticities. The three curves indicates how a constant quantity of fishing effort may be produced by three different production technologies, where the substitution elasticity is zero (the red curve), between zero and infinity (the blue curve) or infinity (the green curve).
As discussed in Section 2.2 fishing effort may be measured in many ways. The labour involved could be measured in terms of number of hours the fishers spend on fishing. It may also be relevant to include the time spent on preparing for fishing and for finishing it. Of practical reasons, we prefer to measure labour and capital by the same units. The natural unit to use is value in monetary units per unit. The value of labour and capital used in the production then gives the total cost of the produced fishing effort.
2.4 Introduction to elasticities Economic theory often makes use of different marginal measures - average values, derivatives and elasticities - all of them reflecting per unit changes. For a function f(x), the per unit values are f(x) • the average value x • the derivative with respect to the variable x, f 0(x) 0 f(x) • the elasticity with respect to the variable x, f (x)/ x
As seen from this listing the elasticity is the ratio between the two other, the derivative divided by the average value. As a ratio the elasticity is unit less, which in particular is an advantage when f(x) can be measured in different units. The interpretation of the elasticity of f(x) then is the percentage change in f(x) when x is changed by 1%. 14 Chapter 2. Fishing effort
2.5 The Cobb-Douglas function Equation 2.1 may have different mathematical expressions depending on the properties of the production processes. The Cobb-Douglas function is often used to describe a production process, assuming unit elasticity of substitution and an elasticity of scale equal 1. We will return to these expression later, first we will introduce the Cobb-Douglas production function.
Let and A be positive constant terms and the Cobb-Douglas production is expressed by
E(L,K) = A · Lα · K1−α (2.2)
where A and α are non-negative constants, and 0 ≤ α ≤ 1. The Mathematica implementation of equation 2.2 is given by In[1]1 in Code box 2.5.1 below where also the output elasticities are found. Code box 2.5.1 — Output elasticities in the Cobb-Douglas function. Defining the Cobb-Douglas function in Mathematica:a
In[1]:= cd[l_, k_] := A * l^α * k^(1 - α)
Now find the output elasticity with respect of labour (l):
In[2]:= D[cd[l, k], l] * l / cd[l, k]
Out[2]= α
and the output elasticity with respect of capital (k):
In[3]:= D[cd[l, k], k] * k / cd[l, k]
Out[3]= 1 - α
aSince all internal Mathematica commands start with a capital letter we prefer to use lower case letters in our variables, to avoid confusion. In this code the fishing effort symbol (E) in equation 2.2 is replaced by cd, indicating that it is a Cobb-Douglas equation.
The cost originating from the capital used is twofold: The market value of the fishing gears and other materials used in the fishing operation (c), plus the value of the forgone benefits by spending the capital on materials for fishing (co) and not elsewhere. The latter is often referred to as the opportunity cost of capital.
The cost of labour could be calculated in different ways. If the labour is hired, the labour cost is the wages paid (w) and the opportunity cost of the capital spend on labour (lo). For an independent fisher, spending the time on fishing, the labour cost is the forgone income by utilising the labour achievement in the best alternative placement (co).
The three fishing technologies mentioned above (angling, long line and hand line fish- ing) use different mixes of labour and capital in their production of fishing effort.
1Mathematica inputs are numbered in this way and can be referred to in various ways, for example by the expression ’%1’. 2.6 The CES function 15
2.6 The CES function The CES function gives a more general formulation of equation 2.1. The abbreviation CES means Constant Elasticity of Substitution and figure 2.2 shows the two extremes of an infinitely large elasticity of substitution (green) and zero elasticity of substitution (red).
The Cobb-Douglas function is a special case of the CES function, with a constant elasticity of substitution equal one (unit elasticity of substitution). The blue curve in figure 2.2 shows a Cobb-Douglas curve while all three curves are special cases of a CES func- tion. Code box 2.6.1 — Output elasticities in the CES function. Defining the CES function in Mathematica:
In[1]:= ces[l_, k_] := (α l^r + (1 - α) k^r)^(1/r))
where r is a parameter related to the elasticity of substitution. The elasticity of substitution is η = 1/(1 − r). Now find the output elasticity with respect of labour (l):
In[2]:= D[ces[l, k], l] * l / ces[l, k]
lr α Out[2]= kr (1-α) + lr α and the output elasticity with respect of capital (k):
In[3]:= D[ces[l, k], k] * k / ces[l, k]
kr (1-α) Out[3]= kr (1-α) + lr α Let us make 3D plots of the three cases indicated in figure 2.2, when the elasticity of substitution (η) equals ∞, 1 and 0 respectively. In the plots the input factors, labour and capital, are represented by the two horizontal axes while the output is measured upwards.
For η = ∞ (perfect elasticity of substitution):
In[4]:= Plot3D[ces[l, k] /. {r -> 1, α -> 1/2}, {k, 0, 1}, {l, 0, 1}, MeshFunctions -> {#3&}, Mesh -> {Transpose[ {Range[5]/6,{#,#,Directive[Thickness[.02], Green],#,#}} ]& @Black}]
Out[4]=
For η = 1 (the Cobb-Douglas function): 16 Chapter 2. Fishing effort
In[5]:= Plot3D[Limit[ces[l, k]/.{α->1/2},{r->0}], {k, 0, 1}, {l, 0, 1}, MeshFunctions -> {#3&}, Mesh -> {Transpose[{Range[5]/6, {#,#,Directive[Thickness[.02],Lighter@Lighter@Blue],#,#}} ]& @Black}]
Out[5]=
For η = 0 (the Leontief function):
In[6]:= Plot3D[ ces[l, k]/.{r->1000000000000, α->1/2}, {k, 0, 1}, {l, 0, 1}, MeshFunctions -> {#3&}, Mesh -> {Transpose[ {Range[5]/6,{#,#,Directive[Thickness[.02], Red],#,#}} ]& @Black}]
Out[6]=
In each of the three plots above a constant value of output is indicated by colour according to the colours used in figure 2.2, green (η = ∞), blue (η = 1) and red (η = 0).
2.7 Elasticity of scale The principle of elasticises is presented in section 2.4 and output elasticities are found for the Cobb-Douglas function (Mathematica code 2.5.1) and the CES function (Mathematica code 2.6.1). The interpretations of the output elasticities are straightforward: When one of the input factors, l or k, changes (increases or decreases) by one percent, the production output is changed accordingly by the percentage equal the output elasticity of the input factor.
According to the results in Mathematica code 2.5.1 the output elasticity of labour in the Cobb-Douglas function is α and the output elasticity of capital is 1 − α. If both input factors are changed simultaneously by one percent, the total change of output equals the sum of the two output elasticities. For the Cobb-Douglas equation it is easy to see that α + (1 − α) = 1. This sum is usually referred to as the elasticity of scale (or returns to scale), indicating the effect up- and down-scaling have on the produced quantity. 2.7 Elasticity of scale 17
Also for the case of the CES function the elasticity of scale equals one in this case. Here it is convenient to employ Mathematica to find the solution.
In[7]:= Simplify[D[ces[l, k], l]*l/ces[l, k] + D[ces[l, k], k]*k/ces[l, k]]
Out[7]= 1
The elasticity of scale is usually assumed to equal one, meaning that there not is economics of scale in the production. This may be considered as the normal case but is certainly always the case. Both the Cobb-Douglas function and the CES function may be adjusted to accommodate scale elasticities different from one. A simple and general formulation of the first function is
E(L,K) = A · Lα · Kβ (2.3) where the elasticity of scale is = α + β.
3. Fish catch production
3.1 Catch production In section 2.1 necessary requirements of producing fishing effort were discussed. It is however not sufficient to know how to catch fish, fish stock resources also need to be available for catch. In economic terms we would say that the fish stock is an essen- tial input factor in the production of catch. The other essential factor in the production of catch is the fishers’ fishing aptitude which in the following referred to as the fishing effort.
Catch production is a production process following the principle of technological effi- cient production discussed in chapter 2 where some well-known production functions were introduced. In this chapter we will discuss further how to implement the Cobb-Douglas function (equation 2.3) to describe fish catch production.
In chapter 2 we have already discussed how fishing effort (E) is produced by the in- put factors labour (L) and capital (K). Now we consider a production process where fishing effort as an input factor in the output is fish harvest. Fishing effort alone is not sufficient to produce harvest of fish, it is also necessary to have access to a fish stock resource. Therefore the available fish stock also is an input factor in the production of fish harvest
In chapter 4 we will discuss further the properties of the fish stock, here we only consider the fish stock in terms of available biomass as one of the two input factors to produce fish catches.
3.2 Catch as a method of stock assessment Stock assessment is the core foundation of modern fisheries management. Time series of stock assessments provides information about growth potential, variability and possible exploitation levels. 20 Chapter 3. Fish catch production
The simplest stock assessment method is to compare the amount of catch per unit of fishing intensity (catch per unit of effort, CPUE) between different areas or periods. The highest CPUE is believed to reflect the highest abundance.
This is however not a method to estimate the real number of fish in an area at a point in time but merely a relative measure to rank different observations. However, in order to tell if the fish abundance is increasing or decreasing it is sufficient to measure the relative changes.
The indexes by which we measure relative changes (e.g. CPUE) are regarded as stock estimates. In fact, this is the case also for the more sophisticated stock assessment methods utilised in data rich single species fisheries as for example the Northeast Arctic cod fishery. In this fishery there exists high resolution catch data, annual survey data and a number of scientific studies contributing in enriching the stock assessment methodology and tuning processes used to evaluate the state of the stock at any point in time.
Let us move back to the basic observation which we may label the CPUE methodol- ogy, where a large catch with a given effort signalise a larger stock abundance than a small catch with the same effort at another point in time. In its most rudimentary form the CPUE methodology assumes a linear catch function:
H(E,X) = q · E · X (3.1) where H is the harvest produced by the fishing effort, E, and the stock abundance (mea- sured in biomass), X. q is often referred to as the catchability coefficient, a scaling parameter also reflecting the technological property of the fishing gear in use.
As seen in section 2.7 It is easy to see from equation 3.1 that catch per unit of effort (CPUE) is linear to the stock index q · X:
CPUE = H(E,X)/E = q · X (3.2)
Code box 3.2.1 — Simple stock assessment by CPUE calculations. Assume that we have access to time series of catch data and standardised fishing effort over a period of ten years. The catch time series is:
In[1]:= catch = { 2384, 2361, 1586, 1889, 1766, 2456, 1068, 1905, 1425, 1957};
The effort was 100 the first year and increased by 10 every year after. The CPUE development over time is seen by plotting catch per effort for each year
In[2]:= ListLinePlot[catch/Table[100 + 10*t, {t, 0, 9}], Mesh -> All, Frame->True,PlotTheme->"Detailed",FrameLabel->{"Year","CPUE"}] 3.2 Catch as a method of stock assessment 21
25
20
15
Out[2]= CPUE 10
5
0 0 2 4 6 8 10
Year
We observe a down-sloping trend in the CPUE measures and assume a linear model
In[3]:= model1 = LinearModelFit[ Transpose[{#, catch/#} &@Table[100 + 10*i, {i, 0, 9}]], x, x]
Out[3]= FittedModel 35.4802- 0.149425 x
We see that about 67% of the variation is explained by the linear model
In[4]:= model1["RSquared"]
Out[4]= 0.675255
and we retrieve the analysis of variance by
In[5]:= model1["ANOVATable"]
DF SS MS F-Statistic P-Value
Out[5]= x 1 184.204 184.204 16.6347 0.00354115 Error 8 88.5878 11.0735 Total 9 272.792
The results of the linear regression is plotted together with the CPUE observation versus fishing effort.
In[6]:= Show[{Plot[model1[e], {e, 0, 240}, PlotStyle -> Dashed], ListLinePlot[Transpose[{#, catch/#} &@Table[100+10*i, {i,0,9}]], Mesh -> All]}, Frame -> True, PlotRangePadding -> None, PlotRange -> {0, All}, FrameLabel -> {"Effort", "CPUE"}] 22 Chapter 3. Fish catch production
35
30
25
20
Out[6]= CPUE 15
10
5
0 0 50 100 150 200
Effort
Multiplying the CPUE values above with effort gives the catch and the linear regression now describes a parabolic curve through the origin.
In[7]:= Show[{Plot[model1[e]*e, {e, 0, 240}, PlotStyle -> Dashed], ListLinePlot[Transpose[{Table[100 + 10*i, {i, 0, 9}], catch}], Mesh -> All]}, Frame -> True, PlotRangePadding -> None, PlotRange -> {0, 2600}, FrameLabel -> {"Effort", "Catch"}]
2500
2000
1500
Out[7]= Catch 1000
500
0 0 50 100 150 200 Effort
If we had an accurate estimate of the value of q, we could actually determine the stock biomass in nature, X. In most cases it is however sufficient to know if the stock is growing or declining to make non-critical management decisions. Then the CPUE-measure is sufficient; given that equation 3.1 holds and that we are able to measure catch (H) and effort (E) correctly. The latter is a major problem which is dealt with other places in this document. Here we will discuss further the methodology, given that accurate catch and effort observations exist.
Measurement of fishing effort in a consistent manner is a challenge which is not eas- ily solved. There are no standard methods of standardising effort in a heterogeneous fleet at a point in time. In addition, fleet efficiency changes (typically increases) over time, creating a systematic error when comparing fishing effort measures in different time periods.
In principle catch quantities are easier to measure than fishing effort. It demands however a system of retrieving such information without hidden, manipulated or illegal catches 3.3 Cobb-Douglas catch production 23
disturbing the data samples.
3.3 Cobb-Douglas catch production Catch is produced by two input factors, fishing effort (E) and stock biomass (X) as indicated in equation 3.1. Now assume a more general catch equation following the pattern of a Cobb-Douglas production function (equation 2.3)
H(E,X) = q · Eα · Xβ (3.3)
As seen from equation 3.3 this is equivalent to equation 3.1 when α = β = 1. There may be reasons to expect α = 1 and 0 < β < 1 (Hannesson, 1983[5]; Eide et al., 2003[3]). The value of β depends both on biological properties of the fish stock in question (typically β will be lower for schooling species than for non-schooling species) and the properties of the harvesting technology (gillnetting has for example higher β-values than longlining).
On the basis of the reasoning above let us assume an alternative catch equation where β = 1/2: √ H(E,X) = q · E · X (3.4)
3.4 Properties of fishing gears In chapter 2 a number of different fishing gears are mentioned. Some gears are fishing randomly fish passing through the area (some traps and gill nets), some gears aim to attract fish to swim into traps or onto hooks where they are captured (fish pots, long line, hand line, etc.), while other fishing gears actively move towards the fish to capture it (trawl, purse seine, Danish seine, etc.).
3.5 Exercises Exercise 3.1 Why increases the production more rapidly in the lower-right figure in
Code box 6.1.2 than in the upper-right figure?
4. Population dynamics
4.1 Basic principles A group of individuals of the same species living in a specific geographically area is often referred to as a population. Populations may be separated into smaller sub-populations (often referred to as stocks) in minor geographical areas. Population dynamics describes how a population or stock grows due to the genetic properties of the population/stock and the environmental constraints within the distribution area. In the following we will not differentiate between populations and stocks.
Population growth is the net effect of individual growth, recruitment and mortality. All these factors depend both of the biological properties of the species and of the physical and biological environment. The size of a population could be measured in numbers of individuals (the common measure in the human population) or in terms of weight (the common measure of fish stocks). The latter is often referred to as the stock’s biomass, which is the total weight of the stock in nature.
The science of population dynamics originates from the same public discussions lead- ing to the economic discipline and mathematical demographics at the end of the eighteenth century. Several of the demographic models we will discuss in this chapter originates from that period, but at that time the focus was on the growth of the human population.
4.2 Classical surplus growth models Of course people had been interested in and fascinated by population growth long time before the eighteenth century. One example is the Fibonacci numbers, described by an Italian mathematician (Leonardo of Pisa Bonacci, about 1170 – about 1250)) during the first years of the thirteenth century (the series has however been known from ancient times).
The Fibonacci numbers describes biological growth starting with a newborn pair of rabbits 26 Chapter 4. Population dynamics
(1). At the end of the first month they mate and a new pair is born at the end of the second month. Then there are two pairs. The first pair gives birth to a new pair every month (mortality is not existing in this model), while the second pair starts mating as the first pair after their first month. The Fibonacci series then gives the monthly number of pairs of rabbits.
Code box 4.2.1 — Fibonacci numbers. In Mathematica you have an internal function listing the Fibonacci numbers:
In[1]:= Table[Fibonacci[n], n, 20]
Out[1]= {1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987, 1597, 2584, 4181, 6765}
In[2]:= ListLinePlot[%1, Mesh -> All, PlotRange -> All, Filling -> Axis]
7000
6000
5000
4000 Out[2]= 3000
2000
1000
5 10 15 20
Now have a look at the ratio between two and two consecutive numbers
In[3]:= Fibonacci[# + 1]/Fibonacci[#] & /@ Range[20]
3 5 8 13 21 34 55 89 144 233 377 610 Out[3]= 1, 2, , , , , , , , , , , , , 2 3 5 8 13 21 34 55 89 144 233 377 987 1597 2584 4181 6765 10946o , , , , , 610 987 1597 2584 4181 6765 The numerical values reveal that the ratios soon approach the same value while moving upwards in the series.
In[4]:= N[%]
Out[4]= {1., 2., 1.5, 1.66667, 1.6, 1.625, 1.61538, 1.61905, 1.61765, 1.61818, 1.61798, 1.61806, 1.61803, 1.61804, 1.61803, 1.61803, 1.61803, 1.61803, 1.61803, 1.61803}
Let us look at the numerical value of the built in function GoldenRatio, and we obtain the same number.
In[5]:= GoldenRatio // N 4.2 Classical surplus growth models 27
Out[5]= 1.61803
In[6]:= Show[{ListLinePlot[%3, PlotRange -> All, Mesh -> All], Plot[GoldenRatio, {n, 0, 20}, PlotStyle -> Red]}]
2.0
1.5
Out[6]= 1.0
0.5
5 10 15 20
The blue curve (Out[3] falls together with the red curve (the golden ratio, Out[4]). The golden ratio therefore could be obtained directly from the Fibonacci numbers:
In[7]:= Limit[Fibonacci[n + 1]/Fibonacci[n], n -> Infinity]
1 √ Out[7]= 1 + 5 2 Let us complete this session with the Fibonacci numbers by looking at one of many examples of how these numbers produce beautiful patterns in the nature. Here we use the golden ratio to draw patterns found in sunflower heads
In[8]:= Graphics[{Blend[{Yellow, Orange}, 1/4], Disk[{0, 0}, 23], Brown, PointSize[.02], Point[Table[Sqrt[n] { Cos[2 Pi n GoldenRatio], Sin[2 Pi n GoldenRatio]}, {n, 500}]]}]
Out[8]=
An interesting feature of the series is that each number is equal the sum of the two previous numbers. At first sight the series may appear to be rather randomly constructed but it is following a strict rule where the population of rabbits after only a few months reach 28 Chapter 4. Population dynamics astronomical values. The series is a simple recruitment model and does not include growth in biomass and mortality. It demonstrates however the exponential power of population growth which was also discovered in European countries during the eighteenth century.
The British scholar Thomas Robert Malthus (1766 - 1834) claimed that the exponen- tial increase in the human populations throughout Europe in the eighteenth century eventually would be repressed at a maximum level (the level of subsistence) where the human population would suffer at the edge of natures capacity level of sustaining the human population.
Malthus’ ideas became very influential but also criticised by many. The idea that there existed an upper limit for how large a human population could be, and that this limit was related to environmental constraints (e.g. food), was shared by most of the critics.
Code box 4.2.2 — Gompertz’ population growth model. The Gompertz equation 4.1 is solved by Mathematica while providing a value for the initial stock biomass (X0, implemented below as x0). The solution includes a message indicating that other solutions may exist in the general case of such
In[1]:= DSolve[x’[t] == -r x[t] Log[x[t]/k],x[0] == x0, x[t], t][[1, 1]]
Solve::ifun: Inverse functions are being used by Solve, so some solutions may not be found; use Reduce for complete solution information. >>
-r t x0e Out[1]= x[t] → k k
Plotting equation 4.1 (left below) and its solution (right below) for some given parameter values (r = 0.5, K = 1000 and X0 = 10)
In[2]:= GraphicsRow[{Plot[-.5 x Log[x/1000], {x, 0, 1000}], Plot[1000 (10/1000)^Exp[-.5 t], {t, 0, 12}]}]
1000 150 800
100 600 Out[2]= 400 50 200
200 400 600 800 1000 2 4 6 8 10 12
Benjamin Gompertz (1779 – 1865) published in 1825[4] a mathematical demographic model. When replacing the variable (numbers of peoples) with a stock’s biomass (X), the Gompertz’ growth equation gives the time derivative of the stock biomass
dX(t) X(t) X˙ (t) = = −r · X(t) · ln (4.1) dt K where r is a growth rate while K is the environmental carrying capacity level. Since the time derivative is the per unit of time increment in the stock biomass, it provides a straight 4.2 Classical surplus growth models 29 forward biological interpretation on the speed by which the stock grows towards its natural equilibrium K. As seen from the plot of equation 4.1 in Code box 4.2.2, the point of maximum growth is placed to the left of K/2.
Another demographic model, published only a few years after the Gompertz model, and with an even greater scientific impact, was suggested by Pierre François Verhulst (1804 - 1849) in 1838 [8]. It is commonly referred to as the logistic growth equation and in line with the expression above it may be written as:
X(t) X˙ (t) = r · X(t) 1 − (4.2) K
As seen from equation 4.2 (and illustrated in Code box 4.2.3) the time derivative of the logistic growth describes a parabolic curve. The inflection point of X(t) therefore is found for X(t) = K/2. After Raymond Pearls introduction of the logistic growth curve as a biomass surplus growth model it has become the most common surplus growth model used in mathematical biology and fisheries economics.
Code box 4.2.3 — Verhults population growth model. The Verhulst equation 4.2 is solved by Mathematica while providing a value for the initial stock biomass (X0, implemented below as x0). The solution includes a message indicating that other solutions may exist in the general case of such
In[1]:= DSolve[x’[t] == r x[t](1 - x[t]/k), x[0] == x0, x[t], t][[1, 1]]
Solve::ifun: Inverse functions are being used by Solve, so some solutions may not be found; use Reduce for complete solution information. >>
ert k x0 Out[1]= x[t] → k - x0 + ert x0 Plotting equation 4.2 (left below) and its solution (right below) for some given parameter values (r = 0.5, K = 1000 and X0 = 10)
In[2]:= GraphicsRow[{Plot[.5 x (1 - x/1000), {x, 0, 1000}], Plot[(Exp[.5t]1000*10)/(1000-10+Exp[.5t] 10), {t, 0, 20}]}]
120 1000 100 800 80 600 Out[2]= 60 400 40 20 200
200 400 600 800 1000 5 10 15 20
In 1959 F. J. Richards (1901 – 1965) published a model in which the two previously presented models (equations 4.1 and 4.2) are special cases, referring to it as "a flexible growth function for empirical use"[6]. In line with the previous equations we express 30 Chapter 4. Population dynamics
Richards function in this way
X(t)m−1 X˙ (t) = r · X(t) 1 − , (4.3) K introducing a third parameter (m), which moves the inflection point of the growth up- wards and downwards depending on its value. It is easy to see that equation 4.3 is equal equation 4.2 for m = 2. It is also possible to show that equation 4.3 actually approaches equation 4.1 when m approaches one. Then both the Gompertz growth and the Verhulst growth equations are special cases of the Richards growth in equation 4.3. The graph shown in Code box 4.2.4 illustrates how the m-parameter in the Richards growth equation determines if the growth curve should be squeezed to the left or the right.
Code box 4.2.4 — Surplus growth models. This session makes use of the package PopulationGrowth (freely available at http://www.maremacentre.com/econmult). If the package is found by Mathematica in its file system, it is loaded by the command:
In[1]:= Needs["EconMult‘PopulationGrowth‘"]
This lists four surplus production models available in the package:
In[2]:= Grid[Text /@ {#, Notation@SurplusProduction[UseMSY -> False, GrowthModel -> #], SimplifyNotation@SurplusProduction[GrowthModel -> #]} & /@ $SurplusProductionModels, Frame -> All, Alignment -> Left]
VerhulstSchaefer r X1- X 4 MSY X(K-X) K K2 GompertzFox-r X log X ⅇ MSY X(log(K)-log(X)) K K
X m m-1 MSYX-K Out[2]= RichardsPellaTomlinson r X1- X K K 1 m K m 1-m -m 1-m
b b 1 X b+1 b +1 (b+1) MSY K -X X -1 b K QuasiBevertonHolt r X1- b+1 K K
In[3]:= Show[(Plot[SurplusProduction[ GrowthModel -> RichardsPellaTomlinson, CurrentBiomass -> x, UseMSY -> True, MaximumSustainableYield -> 100, CatchabilityCoefficient -> 1, BiomassMaximum -> 1000, RichardsPellaTomlinsonParameter -> #], {x, 0, 1000}, PlotRange->{0, 100}, Frame->True, FrameLabel -> {"Stock size", "Natural growth per unit of time"}, PlotRangePadding -> None ] &) /@ Table[Exp[.001 + .1*i^3] - 1, {i, 0, 4, .5}]] 4.3 Depensatory growth 31
100
80
60
Out[3]= 40
Natural growth per unit of time 20
0 0 200 400 600 800 1000 Stock size
4.3 Depensatory growth The models introduced in section 4.2 above are all included in the class of compensatory growth models. The term compensatory refers to the fact that in these models the relative growth rate (X˙ (t)/X(t)) is increasing for decreasing stock biomass levels. Biomass losses leads to an increasing compensation per unit of biomass.
This kind of compensating behaviour in the stock may however not always be the case in real life. The relative growth rate may increase by decreasing stock biomass down to a certain level where the relative growth rate also decreases. It may even turn negative in some cases. These cases are referred to as critical depensation, defining a critical biomass level below which the stock will go extinct.
A depensatory growth model may be specified on the basis of the logistic equation 4.2 by adding a term including a depensation parameter D:
X(t) X(t) − D X˙ (t) = r · X(t) 1 − · K K − D which gives
r · X(t) · K − X(t) · X(t) − D X˙ (t) = (4.4) K · K − D
If the depensation parameter D is negative the depensation is non-critical, while the interpretation of a positive depensation parameter is the critically low biomass level. By the use of Mathematica it is easy to show that D = −∞ restores equation 4.2:
In[1]:= Limit[(r x (k - x) (x - d))/(k (k - d)), d -> -Infinity]
r (k - x) x Out[1]= k 32 Chapter 4. Population dynamics
Code box 4.3.1 — Depensatory growth. Depensatory growth as expressed in equation 4.4 includes both critical and non-critical depensation levels. The plot of equation 4.4 (below) for different D-values illustrates this. The chosen colours goes from non-critical (blue) to critical (red) depensation levels.
In[1]:= Show[Plot[r x(k-x)(x-d))/(k(k-d)) /. {r->.5, k->1000, d->#}, {x, 0, 1000}, PlotStyle -> Hue[.8-Abs[Log[2 + #/3000]]]] & /@ {-50000, -3000, -1000, -300, 0, 200, 400}, PlotRange -> All, AxesLabel -> {"X", "dX/dt"}]
dX/dt
100
Out[1]= 50
X 200 400 600 800 1000
Two of the D-values above are positive and hence defining critical biomass levels, respec- tively for stock sizes of 200 and 400. The blue curve (D = −50,000) is approaching the logistic growth function (obtained when D = −∞).
The plot below shows the relative growth (or average growth; surplus production per unit of biomass) for the D-values included in the plot above.
In[1]:= Show[Plot[r x(k-x)(x-d))/(k(k-d))/x /. {r->.5, k->1000, d->#}, {x, 0, 1000}, PlotStyle -> Hue[.8-Abs[Log[2 + #/3000]]]] & /@ {-50000, -3000, -1000, -300, 0, 200, 400}, PlotRange -> All, AxesLabel -> {"X", "(dX/dt)/X"}]
(dX/dt)/X
0.4
0.2 Out[1]=
X 200 400 600 800 1000
-0.2
We see that the two orange or red curves, representing the two cases of critical depensa- tion, becomes negative at sufficiently low D-values. Other cases also indicate depensation 4.3 Depensatory growth 33
but not all of them. The blue and magenta coloured curves obviously shows an increasing trend by decreasing D-values in the whole range of D-values. This also seems to be the case for the green curve.
The plot below gives the marginal values of the relative growth, with respect of stock biomass X. This plot confirms that the blue, magenta and green curves indeed belong to the class of compensatory growth models, while the other cases belong to the class of depensatory growth models. We base this conclusion on the fact that the first three cases do not enter the positive region of marginal relative growth. Among the depensatory growth cases the orange (red) curves represent critical depensatory growth, while the other two (the yellow curves) represent depensatory growth where no critical biomass values above zero are found.
In[1]:= Show[Plot[D[r x(1-x/k)(x-d)/(k-d)/x, x] /. {r->.5,k->1000,d->#, x->y}, {y, 0, 1000}, PlotStyle->Hue[.8-Abs[Log[2+#/3000]]]]& /@ {-50000, -3000, -1000, -300, 0, 200, 400}, PlotRange -> All, AxesOrigin -> {0, 0}, AxesLabel -> {"X", "-(dX/dt)/X2"}]
-(dX/dt)/X2
0.0010
Out[1]= 0.0005
X 200 400 600 800 1000
-0.0005
The curves intersections with the horizontal axis corresponds to the maximum values of the relative growth. While the green curve reach zero at D = 0, the magenta and blue curves do not reach positive values for D ≥ 0.
Analytically it is possible to prove that the range of D-values of compensatory growth is from −∞ to −K, non-critical depensatory growth goes from −K to 0, while the critical area lays in the interval of D-values between 0 and K, being the maximum biomass level.
Most all examples provided in this text book assume compensatory growth. Shifting to depensatory growth may severely alter the conclusions of the compensatory growth models. If the stock’s ability to strive for increased growth per biomass unit with declining stock size is weakened, consequences of overfishing may be much more severe.
In the next section age structure models are introduced. Walters et al. (2008)[9] discusses the relationship between depensatory growth in surplus production models versus age structured models and claims that depensation is more commonly included in age structured models. 34 Chapter 4. Population dynamics
4.4 Age structured model Age structured population models decompose the population biomass to different age components, each of them being the product of average individual weight in the age group and number of individuals. Changes in age composition of the stock will affect the stock biomass development in ways not necessarily grasped by the surplus production models presented in section 4.2.
Age structured models are often referred to as cohort models, a more general term which allows the stock to be structured in other ways that by year classes, which is the normal structure. For some species, however, structuring by other time intervals (for example by month) or more aggregated groups (for example mature and immature individuals) is preferable to structuring by year-classes.
In 1934 Karl Ludwig von Bertalanffy (1901 – 1972)[2] launched an individual length growth model for asymptotic growth towards a maximum length (L∞) by increasing age. Let L(t) be the individual length at age t, k the length growth rate and t0 the theoretical age of zero length (L(t0) = 0). The von Bertalanffy equation is then given by