1. FTP Some more mathematical formulation of stock dynamics

2. 2 Purpose of slides Introduce the basics in mathematical representation of population dynamics in some detail: Show how the models are all a a special form of the mass balance equation (Russels equation) Stock production models Cohort based models (generic length/age models) Describe the general pattern observed Derive the mathematical equation Source: Haddon 2001: Chapter 1 & 2 Hilborn and Walters 1992: Chapter 3.4

3. 3 Russel’s mass balance formulation Russels contribution: “.. the sole value of the exact formulation given above is that it distinguishes the separate factors making up gain and loss respectively, and is therefore an aid to clear thinking” (Russel 1931) Recognized that a stock could be divided into animals that were in the fishable stock and those that were entering the fishable stock at any one time (recruitment) Stock biomass has gains: Recruitment and growth Stock biomass has losses: Natural and fishing mortality (catch)

4. 4 Russel’s equation: A mass balance equation B t+1 = B t + R t + G t - M t - Y t B t+1 stock size in weight at start of time t+1 B t stock size in weight at start of time t R t weight of all recruits entering stock at time t Recruits: Young fish “entering” the stock in each time period G t weight increase of fish surviving from t to t+1 M t weight loss of fish that died from t to t+1 Y t weight of fish captured from t to t+1

5. FTP Stock production models

6. 6 Russel’s equation modified Russel’s equation for biomass: B t+1 = B t + R t + G t - M t - Y t In the absence of fishing: B t+1 = B t + R t + G t - M t The two sources of increase are called production: Production = (R t + G t ) Difference between production and natural mortality is called surplus production: Surplus production = (R t + G t ) - M t If the processes of recruitment, growth and natural mortality are constant we can write the Russel equation as: B t+1 = B t + rB t r: intrinsic growth rate Here recruitment, growth and mortality are all lumped into one number r > 0, population will grow r = 0, population remains constant r < 0, population decreases with time

7. 7 Population growth curves Model limitations No population increase or decreases continuously and there seems to be an upper bound due to food / space limitation, predation, competition. This model is thus not realistic to describe long term population change Note this is an exponential model

8. 8 Density dependent model Alter the exponential model by taking into account that population growth rates is a function of population size: r’ = r o – r 1 B where ro is the population growth rate when the population size is small (mathematically NULL) r 1 is a value that scale the rates with population size (B)

9. 9 The logistic model 1 By mathematical derivation, taking into account a linear density dependent effect of birth and death rate, we can expand the exponential model: to the following form: K: The carrying capacity (often written as B max ). r: The intrinsic rate of growth (r): is multiplied by the difference between the current population size and the carrying capacity (K-B t / K).

10. 10 Population trajectory according to logistic model

11. 11 Production as a function of stock size K K /2 Maximum production

12. 12 Functional forms of surplus productions Classic Schaefer (logistic) form: The more general Pella & Tomlinson form: Note: when p=1 the two functional forms are the same If p <> 1, then the density dependence is no longer linear

13. 13 General form of surplus production models The Schaefer production model is related directly to Russels mass-balance formulation: B t+1 = B t + R t + G t - M t – Y t B t+1 = B t + P t – Y t B t+1 = B t + f(B t ) – Y t B t+1 : Biomass in the beginning of year t+1 (or end of t) B t : Biomass in the beginning of year t P t : Surplus production the difference between production (recruitment + growth) and natural mortality f(B t ): Surplus production as a function of biomass in the start of the year t Y t : Biomass (yield) caught during year t

14. 14 Production model: The model and data needs Need a time series of: Total annual catch (Y t ) Index of abundance (CPUE t ) This index is most often obtained from data collected on the total effort in the commercial fisheries. In rare cases independent scientific surveys are used in such analysis.

15. 15 The data and the results The data needed: Catch Index of abundance The results: Get an estimate of MSY Get an estimate of Fmsy Current stock status and fishing mortality The issue: Is the model and are the estimates a true reflection of the population? Are the data informative? Catch Abundance index

16. The reference points from a production model B MSY = K/2 F MSY = r/2 MSY = F MSY * B MSY = r/2 * K/2 = rK/4 E MSY = F MSY /q = (r/2)/q = r/(2q) F t = C t /B t

17. FTP … just some word of caution

18. 18 Model uncertainties: The true shape of the production

19. 19 Model uncertainties: The cpue-biomass relationship

20. 20 Model uncertainties The assumption of stock production (and age based recruitment) models is that: The carrying capacity (K), the production (r) and the MSY is on average constant over a long period of time This assumption is highly debated at present, particularly in relation climatic and anthropogenic changes A paradox: It is sometimes argued that the longer the time series of data available (catch and cpue) the better the model estimates. However, since K and r are likely changing over long time spans the problems of the model assumption of constant r and K is more likely to become violated when we use long time series But short time series do not enough information to obtain the parameter estimates

21. 21 One way trip “One way trip” Increase in effort and decline in CPUE with time A lot of catch and effort series fall under this category. Time Effort Time Catch Time CPUE

22. 22 One way trip 2 Fit# r K q MSYEmsy SS R1.243 10 5 10 10 -5 100,0000.6 10 6 ---- T10.182 10 6 10 10 -7 100,0001.9 10 6 3.82 T20.154 10 6 5 10 -7 150,0004.5 10 6 3.83 T30.138 10 6 2 10 -7 250,0009.1 10 6 3.83 Identical fit to the CPUE series, however: K doubles from fit 1 to 2 to 3 But the sums of squares are more or less the same Thus no information about K and r in the data Thus estimates of MSY and Emsy is very unreliable

23. 23 The importance of perturbation histories 1 “Principle: You can not understand how a stock will respond to exploitation unless the stock has been exploited”. (Walters and Hilborn 1992). Ideally, to get a good fit we need three types of situations: Stock sizeEffortget parameter lowlowr high(  K)lowK (given we know q) high/lowhighq (given we know r) Due to time series nature of stock and fishery development it is virtually impossible to get three such divergent & informative situations

24. 24 One more word of caution Most abused stock-assessment technique! Published applications based on the assumption of equilibrium should be ignored Problem is that equilibrium based models almost always produce workable management advice. Non-equilibrium model fitting may however reveal that there is no information in the data. Latter not useful for management, but it is science Efficiency is likely to increase with time, thus may have: Commercial CPUE indices may not be proportional to abundance. I.e. Is the relationship is truly proportional?

25. 25 Some other draw back of production models The population is treated as a “lump”: Input Total catch by weight Catch per unit effort Output Total biomass estimates We know however that: Fish grow in size with time, and thus pass through a multitude of life history stages Larval Juvenile Adult The number of fish must by nature reduce with time after birth The size composition of the fisheries may change with time

26. FTP Size/age based models (cohort models)

27. 27 Development of an cohort

28. 28 Cod: Length composition in survey, march 2002

29. 29 Cod: Time series of SURVEY length measurements 1992 1993 1994 1995 1996 1997 1998 1999 405060708090 Numbers 102030 Length

30. iCod – annual length distribution in the FISHERY 30120 cm 2001 2002 2003 2004 2005 2006 6090 length

31. 31 Cod: Length and age composition in survey, march 2002 1234567 <-- age (years)

32. 32 Cod: Cohort change in length with age

33. 33 Haddock: Annual catch - 1985-2004 Annual catch range: 20-55 million fishes

34. 34 Catch in numbers - 1985-2004 (millions)

35. 35 Catch by yearclasses Year class 1985 1988 1990 Note exponential decline in catches of older fish

36. 36 Catch by yearclasses on a log scale Year class 1985 1988 1990 The slope  total mortality = Z

37. 37 Catch of one year class: 1985 Proportionally low catch of small fish due to selection pattern & availability Decline in catches of old fish due to decline in numbers

38. 38 Proportional catch of different year classes by age Year classes Proportionally low catch of small fish due to selection pattern & availability Decline in catches of old fish due to decline in numbers

39. 39 Pattern observed The pattern observed in the catches: Often have multiple cohorts in the fisheries at any given time The contribution of different cohorts to the catches is quite variable Variable year class size Initial increase in numbers with size/age Decline in numbers with size and age How do we extract information on the population and exploitation from such data? Use mathematical models

40. 40 The stock equation: Exponential population decline

41. 41 The stock equation: Logarithmic transformation

42. 42 Russel equation modified 1 Russel equation for biomass: B t+1 = B t + R t + G t - M t - Y t Russel equation for numbers: N t+1 = N t + R t - D t - C t N t+1 stock size in numbers at start of time t+1 N t stock size in numbers at start of time t R t number of recruits entering the stock at time t D t number of fish that died from t to t+1 Ctnumber of fish that are caught from t to t+1 What happened to the growth term??

43. 43 Russel equation modified 2 Russel equation for numbers: N t+1 = N t + R t - D t - C t If we are considering ONE cohort only we can drop the recruitment term and have N t+1 = N t - D t – C t If we assume that the total number dying (D t + C t ) are a proportion of those living we have: N t+1 = N t - mN t = N t (1 – m) = sN t m: proportion of fish that die during time interval t to t+1 s: proportion of fish that survive during time interval t to t+1 s+m = 1

44. 44 Russel equation modified 3 The equation: N t+1 = N t (1 – m) = N t s is an exponential model and the discrete version is: i.e. a negative exponential model m & s: proportional coefficients (never bigger than one) Z: instantaneous coefficient (can be bigger than one)

45. 45 After birth there is only death! After hatching the individuals in a cohort can only decline If this does not hold true the stock unit being assessed is wrongly defined

46. 46 The higher the Z the faster the population numbers decline

47. 47 Cohort stock equation for each year Since mortality is not constant through out a cohorts life we work in smaller time steps: N t : Number of fish at age time t N t+1 : Number of fish at time t+1 Z t : instantaneous mortality coefficient over time period t to t+1

48. 48 Separation of fishing and natural mortality Since we are often interested in separating natural and fishing mortality we write: M t : natural mortality at time t F t : fishing mortality at time t We will refer to this equation as the stock equation.

49. 49 The effect of fishing M=0.2

50. 50 Describing the catch 1 The number of fish that die in each time interval is: Substituting with the stock equation we get:

51. 51 Describing the catch 2 The number that die due to fishing mortality is the fraction (F t /Z t ) of the number of fish that die, i.e. C t : The number of fish caught over time t to t+1

52. 52 Describing the catch 3 It can be shown that if we take the average* population size of the period t to t+1 (N bar ) the catch equation becomes: The form of the stock equation is then however more complex Where Ti is the time between t and t i (often 1 year) * More precisely: the integral

53. 53 Stock & catch equation  Russel equation Need to have information on weights (direct measurements or from a growth model) to relate back to the original Russel biomass model: B t+1 = B t - Y t Note that we ignored totally the recruitment term since only dealt with one cohort. that is called a “Depletion model” In age/length based stock models we need to make some strong assumptions about the relationship between stock and recruitment. Wt: Measured directly or obtained from the VBGF

54. 54 Cohort analysis: Minimum measurements needed Used to estimate mortality (Z, F) and abundance (N) Mortality (Z) can be estimated from both age and relative size frequency samples survey or catch data by length length based samples: Need to know growth (K and L inf ) For fishing mortality (F) we need to know M F = Z – M For abundance (N) we also need the total catch removed

55. 55 Age/size structured models Advantages Populations do have age/size structure Basic biological processes are age/size specific Growth Mortality Fecundity The process of fishing is age/size specific Relatively simple to construct mathematically Model assumption not as strict as in e.g. logistic models Disadvantages Sample intensive Data often not available Mostly limited to areas where species diversity is low Have to have knowledge of natural mortality For long term management strategies have to make model assumptions about the relationship between stock and recruitment Often not needed to address the question at hand

56. 56 Measurement uncertainties Measurement uncertainties = observation uncertainties Measurement and sampling error Always in place (irrespective of model), question how big In theory the error is assumed random Bias in the sampling Always the case (irrespective of model), question how big Often very difficult to detect Is the sampling area (catch and cpue) representative of the true biological stock unit? How can one obtain accurate estimates of effort for a particular stock being assessed when the fisheries is actually a mixed, opportunistic fishery? …

57. 57 Quantitative fisheries Recent developments: Getting estimates of assessment uncertainties Does not improved stock assessment estimates rather Realized how poorly our estimates actually are Should we give up trying? No Starting point: Know your fisheries, know your data Back to the basics? New innovated methods?