1: \documentclass{elsart}

2: \usepackage{natbib}

3: \usepackage{graphicx}

4: \usepackage{amsmath, amssymb}

5: \usepackage{cite, citesort}

6: 

7: \hyphenation{}

8: 

9: %%% TeX Start

10: \begin{document}

11: 

12: \begin{frontmatter}

13: 

14: % Title, authors and addresses

15: 

16: \title{Exploitation dynamics of fish stocks}

17: %

18: % Version 2

19: %

20: \author{Hiro-Sato Niwa\thanksref{phone}}

21: \thanks[phone]{Tel.: +81-479-44-5953; fax: +81-479-44-1875.\\

22: \hspace*{3mm} {\it E-mail address:} Hiro.S.Niwa@fra.affrc.go.jp (H.-S. Niwa).}

23: %\author{Hiro-Sato Niwa}

24: %\ead{Hiro.S.Niwa@fra.affrc.go.jp}

25: 

26: \address{Behavioral Ecology Section,

27: National Research Institute of Fisheries Engineering,

28: 7620-7 Hasaki, Kamisu, Ibaraki 314-0408, Japan}

29: 

30: \begin{abstract}

31: I address the question of the fluctuations in fishery landings.

32: Using the fishery statistics time-series collected by the Food and

33:  Agriculture Organization of the United Nations since the early 1950s,

34: I here analyze fishing activities and find two scaling features of

35:  capture fisheries production:

36: (i) the standard deviation of growth rate of the domestically landed

37:  catches decays as a power-law function of country landings with an

38:  exponent of value $0.15$;

39: (ii) the average number of fishers in a country scales to the $0.7$

40:  power of country landings.

41: I show how these socio-ecological patterns may be related, yielding a

42:  scaling relation between these exponents.

43: The predicted scaling relation implies that

44: the width of the annual per capita growth-rate distribution scales to

45:  the $0.2$ power of country landings, i.e.

46:  annual fluctuations in per capita landed catches increase with

47:  increased per capita catches in highly producing countries.

48: Beside the scaling behavior, I report that

49:  fluctuations in the annual domestic landings have increased in the last

50:  30 years,

51:  while the mean of the annual growth rate declined significantly after

52:  1972.

53: \end{abstract}

54: 

55: \begin{keyword}

56: scaling law \sep

57: growth dynamics \sep

58: capture fisheries \sep

59: stock exploitation \sep

60: Fishstat Plus \sep

61: FAO

62: \end{keyword}

63: 

64: \end{frontmatter}

65: 

66: % main text

67: %

68: 

69: \section{Introduction}

70: Fluctuations in fishery landings have been a collective human concern

71: \citep{Botsford-etal97,Pauly-etal02}.

72: Variation in supply (stock size) is a central problem in determining

73: the optimal organization of the fishing industry and in how governments

74: can best regulate fishing effort.

75: Since the industrialization of fishing, fisheries scientists have

76: been subject to intense political pressures:

77: ``Determine the causes of fluctuations in fish catches.''

78: For example,

79: in the mid-1800s the Norwegian parliament asked zoologists

80: Michael Sars and George Ossian Sars

81: to investigate the biology of Norwegian fisheries and to answer why

82: the cod catches from the Lofoten Islands in Northern Norway fluctuated

83: so greatly \citep{Smith94}.

84: 

85: Nonlinear dynamics have been applied to understanding the ups and downs

86: of populations.

87: In the mid-1920s,

88: an Italian biologist Umberto D'Ancona

89: found the increase in predator fish and decrease in prey fish of various

90: species in the Adriatic Sea during the World War I period from a

91: statistical study of the numbers of each species sold on the fish

92: markets of three ports, Fiume, Trieste, and Venice

93: \citep{D'Ancona26,D'Ancona54}.

94: A famous Italian mathematician Vito Volterra,

95: asked by D'Ancona whether

96: he could come up with a mathematical model of predator-prey population

97: dynamics,

98: developed the first and simplest model of predator-prey interactions

99: \citep{Volterra26}.

100: Because an American mathematical biologist Alfred J. Lotka produced

101: the two-species interaction model independently and at about the same

102: time \citep{Lotka25},

103: it is called Lotka-Volterra model.

104: The nonlinear model predicts a cyclical relationship between predator

105: and prey numbers.

106: 

107: Thus, the jagged oscillation in populations was nothing new to

108: ecologist, but before the 1970s, they put most of the patterns down to

109: the stochastic, unaccountable effects:

110: year-to year fluctuations in fish stocks are usually attributed to

111: variability in recruitment, competition, predation, changes in fishing

112: effort, and other sources of so-called environmental noise.

113: \citet{May74,May76} discovered that a density-dependent (i.e. nonlinear),

114: deterministic model of ecology could produce complex, random-looking

115: patterns ``chaos''.

116: \citet{Conklin-Kolberg94}

117: researched the effects of market demand on

118: halibut populations of Pacific Ocean fisheries,

119: and provided a model of chaotic dynamics for the Pacific halibut

120: fishery.

121: \citet{Weisbuch-etal97} studied the influence of biological and

122: socio-economic mechanisms on the amplitude of resource depletion, and

123: showed not only that

124: the depletion is most often due to inertial factors in the dynamics

125: of exploitation, i.e. capital and labor overshoot,

126: but also that capital and labor inertia results into dangerous

127: oscillations of the resource level, eventually finishing with a fishery

128: crash.

129: 

130: A central question in ecology for last decades has been how to quantify

131: the relative importance of stochastic and nonlinear factors for

132: fluctuations in population size

133: \citep{Saether-etal00,Turchin03,Hsieh-etal05}.

134: Characterizing the fluctuation patterns is the major challenge for

135: time series analysis in fisheries and for modeling the process of fish

136: production.

137: One major approach to dealing with ecological complexity

138: is to reduce the system to one or a few species:

139: a number of detailed analysis of specific systems have captured the

140: patterns of fluctuations in fish stocks and catches,

141: including noise-driven fluctuations and chaos

142: \citep{Wilson-etal91,Matsuda-etal91,Matsuda-etal92,Takeuchi-etal92,Higgins-etal97,Bjornstad-etal99,Dixon-etal99,Bjornstad-Grenfell01}.

143: 

144: Inspired by work of \citet{Keitt-Stanley98} and \citet{Keitt-etal02},

145: I here explore an alternative approach to

146: socio-ecological system dynamics, i.e.

147: population dynamics within which fisheries are embedded.

148: From the macro-ecological viewpoint,

149: I analyze the time-series of annual fishery landings (all commercial

150: species items fished inclusive) of 244 countries in total during the

151: period 1950--2002,

152: after excluding non-fish vertebrates such as whales.

153: The data on annual domestically-landed catch are

154: taken from Fishstat Plus \citep{FaoFishStat+},

155: a set of statistical database compiled by the Food and Agriculture

156: Organization (FAO) of the United Nations.

157: FAO is the only one institution that maintains global fisheries

158: statistics.

159: The country-structured time-series of landed catch is a complex source

160: of information.

161: Annual landed catch is the aggregate of a large number of complex

162: interactions between fishers and the stock.

163: Some aspects of landed catch are biological phenomena (biomass of fish

164: vulnerable to the gear) and technical efficiencies (use of machinery

165: power and modern equipments such as echo sounders and fish detection

166: devices);

167: others are economic (market price of fish and the cost of applying

168: effort), legal (regulations applied to fishing effort) and some are

169: behavioral (skill of the fisher targeting the stock and accurate

170: completion of catch records).

171: Besides the fluctuations in fishery landings,

172: I investigate the employment statistics based on the FAO data on the

173: number of commercial and subsistence fishers \citep{NoFishers,SOFIA02}.

174: I report two power-law relationships between the following quantities

175: for the capture fisheries sector:

176: \begin{itemize}

177:  \setlength{\itemsep}{10pt}

178:                  %

179: %

180:  \item[{(i)}] the variability of landings time-series and the country

181: 		 landings

182: %

183:  \item[{(ii)}] the number of fishers in a country and the country

184: 		 landings

185: \end{itemize}

186: In order to understand the scaling structure for the stock exploitation,

187: a relation between these power-law exponents is hypothesized,

188: such as the Widom scaling relation \citep{Widom65a,Widom65b}.

189: 

190: In the paper, symbols $x$ and $n$ are used for the annual

191: domestically-landed catch and the number of fishers in countries,

192: respectively.

193: 

194: 

195: \section{World fisheries trend}

196: Fisheries have recently become a topic for the media with global

197: audiences, as well as for the scientific community

198: [e.g. \citet{Beddington-Kirkwood05,Ehrhardt05,Symes05}],

199: with a focus on the wretched history, the poor current practice, and

200: the future scenarios.

201: For instance,

202: the New Orleans {Times-Picayune}'s eight-part 1996 series by 

203: \citet{McQuaid-etal96}

204: on ``Oceans of Trouble'' won the Pulitzer Prize for public service

205: in 1997,

206: in reporting on threats to the world's fisheries.

207: 

208: In 1950, the newly founded Food and Agriculture Organization of

209: the United Nations began collection of global statistics.

210: Fisheries in the early 1950s were at the onset of a period of

211: extremely rapid growth.

212: Everywhere that industrial-scale fishing (mainly trawling, but also

213: purse seining and long-lining) was introduced, it competed with

214: small-scale, or artisanal fisheries.

215: Throughout the 1950s and 1960s, this huge increase of global fishing

216: effort led to an increase in catches so rapid that their trend

217: exceeded human population growth, encouraging an entire generation of

218: fisheries managers and politicians to believe that launching more boats

219: would automatically lead to higher catches \citep{Pauly-etal02}.

220: 

221: In 2002, fisheries provided direct employment to about 38 million people

222: and accounted for more than 15 percent of the total human consumption of animal

223: protein \citep{SOFIA02}.

224: Globally, first-sale capture-fishery revenues produced about

225: 78 billion US dollars.

226: Total world trade of fish products increased to an export

227: value of 58.2 billion US dollars, continuing the last decade's

228: underlying 4.5 percent annual growth.

229: The world's population has been increasing more quickly than the total

230: food fish supply; as a result the average per capita fish supply has

231: declined.

232: Total production from marine and inland capture

233: fisheries for the world

234: reached 93.2 million tonnes

235: (84.5 million tonnes marine and 8.7 million tonnes inland)

236: [excluding China, 76.6 million tonnes from capture fisheries

237: (70.1 million tonnes marine and 6.5 million tonnes inland)].

238: 

239: Figure~\ref{fig:1}{\sf A} shows the trend of the global harvests with

240: capture fisheries,

241: demonstrating that the global capture production has been exponentially

242: growing.

243: The annual growth rate of global catches

244: (defined as $R \equiv \ln [X(t+1)/X(t)]$,

245: where $X(t)$ and $X(t+1)$ are the total capture production in the years

246: $t$ and $t+1$)

247: changed discontinuously at the critical time 1972,

248: correspondent to the collapse of the Peruvian anchoveta fishery,

249: signaling an abrupt transition of bio-economic regime.

250: The fisheries exploitation has expanded in two-exponential mode, and

251: possibly reached a plateau with no transient period

252: contrary to a logistic growth.

253: 

254: In line with the increase in fisheries production, over the last three

255: decades employment in fisheries and aquaculture has continued to

256: increase in many countries.

257: The world number of fishers and fish farmers has been exponentially

258: (i.e. in a Malthusian mode)

259: growing at annual rate of 3.6 percent

260: (excluding China, 2.8 percent)

261: for the period 1970--2002,

262: depicted in Fig.\ref{fig:1}{\sf B}.

263: The world fishers population has been growing at a faster rate than the

264: earth's human population 

265: (see inserted figure of Fig.\ref{fig:1}{\sf B}).

266: Since 1970 to 2000,

267: the ratio of fishers and fish farmers to the whole population

268: has increased from 0.332 to 0.588 percent

269: (excluding China, 0.349 to 0.490 percent).

270: 

271: Since there is the possibility of China's misreporting \citep{WatsonPauly01},

272: for further analysis, I use the data with the exclusion of China's

273: reported catches

274: (including China does not change the results).

275: 

276: 

277: \section{Scaling of country growth rate}

278: In Fig.\ref{fig:2}, shown are the probability distributions

279: of the annual domestic landings~({\sf A}) and the number of people

280: actively fishing and fish farming in countries~({\sf B}),

281: both which are consistent with a log-normal distribution

282: \begin{equation}

283:  P(x)

284:   =

285:   \frac 1{x\sqrt{2\pi s_x^2}}

286:   \exp

287:   \left[-\frac{\ln^2 (x/\hat{x})}{2s_x^2} \right],

288:   \label{eqn:lognormal-landings}

289: \end{equation}

290: where $\hat{x}$ and $s_x$ denote a geometric mean and a geometric standard

291: deviation of the distribution, respectively.

292: The log-normal relationship remained unchanged for the time

293: period studied.

294: This log-normal distribution of the country landings is suggestive of

295: Gibrat's law on market structure \citep{Gibrat31}.

296: It is interesting to note that,

297: contrary to city sizes (resident population) or US firm sizes (employees

298: and revenue),

299: the country population engaged in fishing and fish farming does not

300: reduce to the Zipf's power-law distribution characterized by the absence

301: of a characteristic size \citep{Zipf47,Axtell01}.

302: However, the conditional probability distributions of fishers population

303: for countries with the same landings show ``fat tails'' with

304: characteristic scales

305: (for details, consult \S \ref{section:stretched-exponentials}).

306: 

307: In the limit of small annual changes in $x$, we can define the relative

308: growth rate as

309: \begin{equation}

310:  r

311:  \equiv

312:  \ln\left[\frac{x(t+1)}{x(t)}\right]

313: \approx

314: \frac{x(t+1)-x(t)}{x(t)},

315: \end{equation}

316: where $x(t+1)$ and $x(t)$ are the annual landings of a country in two

317: consecutive years.

318: I calculate the conditional distribution $p(r|x)$ of annual

319: growth rate for countries with landings $x$.

320: The distribution $p(r|x)$ of the growth rate from 1950 to 1971 is

321: shown in Fig.\ref{fig:3}{\sf A}, depicting a simple tent-shaped form on

322: a semi-logarithmic scale.

323: The distribution is not Gaussian, as expected from the Gibrat model, but

324: rather is exponential:

325: \begin{equation}

326:  p(r|x)

327:   =

328:   \frac 1{\sqrt{2}\sigma_r (x)}

329:   \exp \left(

330: 	-\frac{\sqrt{2} \left| r-\overline{r}(x)\right|}{\sigma_r (x)}

331:       \right),

332:   \label{eqn:laplace}

333: \end{equation}

334: where $\overline{r}(x)$ and $\sigma_r (x)$ are the average growth rate

335: and the standard deviation of $r$ for countries with landings equal to

336: $x$, respectively.

337: An implication of this result is that the distribution of the growth

338: rate has much broader tails than would be expected for a Gaussian

339: distribution.

340: Notice that the fluctuation $\sigma_r$ in growth rate $r$ is

341: independent of $x$.

342: 

343: In Fig.\ref{fig:3}{\sf B}, shown is the empirical conditional

344: probability density of $r$ for countries from 1972 to 2002 with

345: approximately the same landings,

346: suggesting that $p(r|x)$ has the same functional form,

347: consistent with the Laplace distribution [Eq.(\ref{eqn:laplace})],

348: with different widths $\sigma_r (x)$, for all $x$.

349: In these years, contrary to the period of 1950--1971,

350: we expect that the statistical properties of the growth rate $r$ depend

351: on landed catch $x$:

352: the growth-rate distribution depends on the initial value of $x$.

353: 

354: The study of the fluctuations has been shown to give important

355: information regarding the underlying processes responsible for the

356: observed macroscopic behavior.

357: I therefore analyze the fluctuations in the growth rate of the country

358: landings.

359: While different plots for the values of landings fall onto one

360: another for the 1950--1971 period,

361: there is clear dependence of $\sigma_r$ on the value of landed

362: catch for both the periods of 1972--1989 and 1990--2002.

363: Figure~\ref{fig:4}{\sf A} shows that for the periods 1972--2002,

364: $\sigma_r (x)$ statistically decreases with increasing $x$.

365: The width of the growth-rate distribution, $\sigma_r (x)$, scales over

366: three order of magnitude, from landings about $10^{3}$ tonnes up to

367: landings about $10^{6}$ tonnes, as a power law:

368: \begin{equation}

369:  \sigma_r (x) = A x^{-\beta}

370:   \label{eqn:scaling-growth-rate}

371: \end{equation}

372: where $A \approx 0.27$ and $\beta \approx 0.15$ for the both

373: investigated periods

374: (landed value of catch unit in thousand tonnes).

375: 

376: What is remarkable about Eqs.(\ref{eqn:laplace}) and

377: (\ref{eqn:scaling-growth-rate}) is that they govern the growth rates of

378: a diverse set of countries.

379: Fishery labor productivity and capital intensity vary widely among

380: countries.

381: They range not only in their amount of landings but also in what and

382: where they catch:

383: the harvest included in the database is classified according to

384: more than 1000 commercial species items,

385: and according to the inland or marine area.

386: Indeed, the data of growth rate for a wide range of production

387: values plotted in terms of scaled coordinates collapse onto a single

388: curve (Fig.\ref{fig:4}{\sf B}),

389: if the ordinate and the abscissa are chosen as

390: \begin{equation}

391:  p_{\mbox{\scriptsize sc}}

392:   \equiv

393:   \sqrt{2} \sigma_r (x) p(r|x)

394:   \quad\mbox{and}\quad

395:   r_{\mbox{\scriptsize sc}}

396:   \equiv

397:   \sqrt{2}

398:   \left[r-\overline{r}(x)\right]/\sigma_r (x),

399: \end{equation}

400: respectively.

401: Accordingly, the probability density function of growth rate $r$ for a

402: country with $x$ tonnes landings, detrended by the average growth rate

403: $\overline{r}(x)$,

404: should scale as

405: \begin{equation}

406:  p\left(r'|x\right)\mbox{d}r'

407:   =

408:   f_r\left(\frac{r'}{x^{-\beta}}\right)

409:   \mbox{d}\left(\frac{r'}{x^{-\beta}}\right),

410: \end{equation}

411: where the scaling function $f_r$ is exponential, and the detrended growth

412: rate is denoted by $r'\;[\equiv r-\overline{r}(x)]$.

413: 

414: 

415: \subsection{$\Delta t$-year growth rate}

416: We now question whether Eq.(\ref{eqn:laplace}) of landed catches in

417: countries is still valid for the growth rates for longer time scales.

418: Consider the $\Delta t$-year growth rate

419: \begin{eqnarray}

420:  r_{\Delta t}

421:   &\equiv&

422:   \ln \left[\frac{x(t+\Delta t)}{x(t)}\right] \\

423:   &=&

424:    \ln \left[\frac{x(t+1)}{x(t)}\right]

425:    +\ln \left[\frac{x(t+2)}{x(t+1)}\right]

426:    +\cdots

427:    +\ln \left[\frac{x(t+\Delta t)}{x(t+\Delta t -1)}\right] \nonumber.

428: \label{eqn:T-year-growth-rate}

429: \end{eqnarray}

430: Is then $r_{\Delta t}$ for large $\Delta t$ Gaussian?

431: Figure~\ref{fig:5}{\sf A} shows the distributions of

432: growth rates for time scales from 1 year up to 30 years

433: over the entire period (1950--2002) for all countries (excluding China).

434: In Fig.\ref{fig:5}{\sf B} shown are

435: the probability distributions of the normalized growth rate defined as

436: \begin{equation}

437:  g_{\Delta t}

438:   \equiv

439:   \frac{r_{\Delta t}-\overline{r_{\Delta t}}}{\sigma_{\Delta t}},

440: \end{equation}

441: where

442: $\overline{r_{\Delta t}}$ and $\sigma_{\Delta t}$ are the average and

443: the standard deviation of the $\Delta t$-year growth rate.

444: We see that

445: the distributions of $g_{\Delta t}$ are, to a good approximation,

446: symmetric with respect to zero.

447: Figure~\ref{fig:5}{\sf C} shows the positive tails of

448: the complementary cumulative distributions $P(\geq g_{\Delta t})$

449: of the normalized growth

450: rates for the same time scales,

451: compared to the standardized Gaussian distribution with a mean of 0 and

452: a standard deviation of 1.

453: Because of the central limit theorem,

454: the convergence to a Gaussian distribution on longer time scales is

455: observed,

456: as the Gibrat model predicts.

457: 

458: We can further test the convergence to Gaussian behavior by analyzing the

459: moments of the distribution of normalized growth rates.

460: A more global quantitative measure of distributions is in terms of the

461: moments:

462: \begin{equation}

463:  \mu_k

464:   \equiv

465:   \left\langle \left|g_{\Delta t}\right|^k \right\rangle,

466: \end{equation}

467: where $\langle\ldots\rangle$ denotes the average.

468: The fractional moments of the normalized growth rates for

469: $\Delta t = 1$, 10, 20 and 30 years

470: (Fig.\ref{fig:5}{\sf D})

471: show clear indication of convergence to Gaussian behavior with

472: increasing $\Delta t$.

473: However, it is remarkable that

474: they do not satisfyingly approach the Gaussian moments even for large

475: $\Delta t = 30$ years.

476: 

477: Moreover, notice that the distribution of annual ($\Delta t = 1$)

478: growth rates $r_1$ for all countries and all years can be well-described

479: by an exponential (Fig.\ref{fig:5}{\sf A}),

480: \begin{equation}

481:  p(r_1)

482:   =

483:   \frac 1{\sqrt{2}\sigma_1}

484:   \exp \left(

485: 	-\frac{\sqrt{2} \left| r-\overline{r_1}\right|}{\sigma_1}

486:       \right),

487:   \label{eqn:laplace-entire-period}

488: \end{equation}

489: with $\overline{r_1} = 0.04$ and $\sigma_1 = 0.27$,

490: while the conditional probability density of the yearly growth rates for

491: countries with approximately the same landings $x$ depends on $x$

492: [Eqs.(\ref{eqn:laplace}) and (\ref{eqn:scaling-growth-rate}),

493: Figs.\ref{fig:3} and \ref{fig:4}].

494: Using a saddle point approximation \citep{Lee98},

495: we may integrate the conditional Laplace distribution,

496: Eq.(\ref{eqn:laplace}),

497: over a log-normal distribution of the country landings,

498: Eq.(\ref{eqn:lognormal-landings}),

499: and obtain Eq.(\ref{eqn:laplace-entire-period}).

500: 

501: 

502: \section{Scaling structure of the fisheries sector}\label{section:stretched-exponentials}

503: Let us analyze employment data in the fisheries sector

504: in order to gain further insight into the organization of fisheries

505: structure of countries.

506: Figure~\ref{fig:6}{\sf A},

507: showing a scatter plot of the number of fishers versus country

508: landings, suggests that

509: the conditional probability density $\rho(n|x)$ to find a country with

510: total catch $x$ landed by $n$ fishers has a long right tail.

511: I here propose a model for the conditional distributions of fishers

512: population in terms of ``stretched exponentials''

513: \citep{Laherrere-Sornette98}:

514: \begin{equation}

515:  \rho(n|x)

516:   =

517:   \frac{\alpha}{n_0} \left(\frac n{n_0}\right)^{\alpha -1}

518:   \exp \left[-\left(\frac n{n_0}\right)^{\alpha}\right],

519: \end{equation}

520: such shat the complementary cumulative distribution is

521: \begin{equation}

522:  P(\geq n|x)

523:   =

524:   \int_n^{\infty} \rho (n'|x) \mbox{d}n'

525:   =

526:   \exp \left[-\left(\frac n{n_0}\right)^{\alpha}\right],

527:   \label{eqn:stretched-exponential-CCDF}

528: \end{equation}

529: where $n_0$ denotes a reference scale of the distribution,

530: expected to depend on the landings,

531: $n_0 = n_0(x)$.

532: Stretched exponentials are characterized by an exponent $\alpha$ smaller

533: than one.

534: The meaning ``reference'' is that from $n_0$ all moments can be

535: determined, e.g. the means of $n$ and $n^2$ are given by

536: \begin{equation}

537:  \overline{n}

538:   =

539:   n_0 \alpha^{-1} \Gamma (\alpha^{-1})

540:   \label{eqn:moment-str-1}

541: \end{equation}

542: and

543: \begin{equation}

544:  \overline{n^2}

545:   =

546:   2n_0^2 \alpha^{-1} \Gamma (2\alpha^{-1}),

547:   \label{eqn:moment-str-2}

548: \end{equation}

549: respectively,

550: where $\Gamma (q)$ is the gamma function

551: [equal to $(q-1)!$ for $q$ integer].

552: When $\alpha$ is small,

553: $\overline{n}$ will be much larger than $n_0$.

554: The expression~(\ref{eqn:stretched-exponential-CCDF}) of the stretched

555: exponential complementary CDF means that

556: the stretched exponential distribution is qualified by a straight line

557: when plotting the negative log of complementary CDF,

558: $-\ln P(\geq n|x)$,

559: as a function of $n$

560: on a double-logarithmic scale.

561: As shown in Fig.\ref{fig:6}{\sf B},

562: lines appear:

563: the values of the exponent $\alpha$ range from 0.60 to 0.86,

564: numerically evaluated from

565: Eqs.(\ref{eqn:moment-str-1}) and (\ref{eqn:moment-str-2}).

566: Accordingly,

567: the fishers population for countries with the same landings

568: follows a stretched exponential distribution

569: [aka \citet{Weibull51} distribution with exponent lass than one].

570: 

571: Moreover Fig.\ref{fig:6}{\sf A} suggests that

572: the typical (i.e. average) number of fishers increases as a power law

573: with the country landings $x$.

574: We make the hypothesis that the conditional probability density

575: $\rho(n|x)$

576: obey the scaling relation

577: \begin{equation}

578:  \rho (n|x) \mbox{d}n

579:   =

580:   f_n\left(\frac n{x^{\alpha}}\right) 

581:   \mbox{d} \left(\frac n{x^{\alpha}}\right),

582:   \label{eqn:fisher-scaling}

583: \end{equation}

584: where $f_n$ denotes a scaling function.

585: To test the scaling hypothesis in Eq.(\ref{eqn:fisher-scaling}),

586: we plot the scaled quantities

587: \begin{equation}

588:  \rho_{\mbox{\scriptsize sc}}\equiv

589:   x^{\alpha} \rho (n|x)

590:   \quad\mbox{vs}\quad

591:   n_{\mbox{\scriptsize sc}}\equiv

592:   n/{x^{\alpha}},

593:   \label{eqn:scaling-hypothesis-fisher}

594: \end{equation}

595: and observe the data collapse with a exponent $\alpha = 0.71 \pm 0.01$

596: (Fig.\ref{fig:7}).

597: Since the data exhibit scaling, we identify a universal scaling function

598: fitted with a stretched exponential distribution

599: \begin{equation}

600:  f_n(n_{\mbox{\scriptsize sc}})

601:   =

602:   \alpha n_{\mbox{\scriptsize sc}}^{\alpha-1}

603:   \exp \left(-n_{\mbox{\scriptsize sc}}^{\alpha}\right).

604:   \label{eqn:stretched-exponential}

605: \end{equation}

606: Notice that the average number of fishers scales with country

607: landings $x$ as

608: \begin{equation}

609:  \overline{n}(x)

610:   =

611:   \int_0^{\infty} n \rho (n|x) \mbox{d}n

612:   \propto

613:   x^{\alpha}.

614:   \label{eqn:typical-number-fishers}

615: \end{equation}

616: As shown in Fig.\ref{fig:6}{\sf A} this is indeed the

617: case.

618: 

619: The multiplicative cascade model \citep{Frisch-Sornette97} may explain

620: the stretched exponential distribution of fishers:

621: the probability distribution function of

622: the product of $m$ independent identically distributed positive random

623: variables with an exponential-like distribution leads to stretched

624: exponentials

625: [see also \citet{Laherrere-Sornette98}].

626: The reciprocal number of levels in the multiplicative cascade reads the

627: index $\alpha = m^{-1}$.

628: Interpreted within the multiplicative cascade model, the value of the

629: exponent $\alpha\approx 0.7$ corresponds to about one to three generation

630: levels in the generation of a typical fisher.

631: 

632: Notice that the conditional distribution $\rho (n|x)$ reduces to a

633: stretched exponential distribution,

634: while the distribution of the country population engaged in fishing and

635: fish farming is log-normal (Fig.\ref{fig:2}{\sf B}).

636: It is numerically demonstrated (Fig.\ref{fig:8}) that

637: a log-normal distribution of the number of fishers approximately results

638: from a convolution of a stretched exponential distribution $\rho (n|x)$

639: and a log-normal distribution of the country landings,

640: Eq.(\ref{eqn:lognormal-landings}):

641: \begin{eqnarray}

642:  P(n)

643:   & = &

644:   \int_0^{\infty} \rho (n|x) P(x) \mbox{d}x \nonumber\\

645:  & = &

646:   \int_0^{\infty}

647:   \alpha \frac{n^{\alpha-1}}{n_0^{\alpha}}

648:   \exp\left[-\left(\frac n{n_0}\right)^{\alpha}\right]

649:   \ 

650:   \frac 1{x\sqrt{2\pi s_x^2}}

651:   \exp

652:   \left[-\frac{\ln^2 (x/\hat{x})}{2s_x^2} \right]

653:   \mbox{d}x

654: \end{eqnarray}

655: with

656: $n_0 \propto x^{\alpha}$.

657: 

658: 

659: \section{Modeling scaling behavior in the exploitation dynamics}

660: Total catch of a country is made up of individual fisher activity in the

661: country.

662: Fisheries management ranges along a continuum from community to

663: government as sole managers of the resource, and the field of impact of

664: their decisions can be local, regional, provincial, or national.

665: Fishers get involved in their respective management arrangements in

666: a country.

667: 

668: We now consider the following two limiting cases.

669: \begin{itemize}

670:  \setlength{\itemsep}{10pt}

671:                  %

672: %

673:  \item[{(i)}]

674: There are strong correlations between the fishers in a country,

675:  following that the growth dynamics are indistinguishable from the

676:  dynamics of structureless organizations.

677: As a result, there is no country-size dependence of the growth

678:  fluctuation $\sigma_r$, following $\beta = 0$.

679: %

680:  \item[{(ii)}]

681: The annual landed catches by individual fishers change independently of

682:  one another.

683: Then, if the number of fishers in a country were proportional to its

684:  domestic fishery product, the growth fluctuations as a function of

685:  domestically-landed catch would decay as a power law with an exponent

686:  $\beta = 0.5$

687:  \citep{buldyrev-etal97}.

688: This is because the standard deviation of the sum of $n$ independent

689:  quantities grows like $\sqrt{n}$ (i.e. catches by $n$ fishers).

690: %

691: \end{itemize}

692: 

693: As to the stock status during the period 1950--1971,

694: there was much room for continued expansion of capture fisheries

695: \citep{Grainger-Garcia96}.

696: The fishing policy those days

697: would have been virtually that

698: ``{Strive for as much fish as possible from the resources.}''

699: All the decisions in a country would have been perfectly coordinated

700: as if they had been all dictated by a single ``boss''.

701: The state of stock exploitation is then expected to have corresponded to

702: the case (i):

703: annual catches by individual fishers in a country were correlated

704: strongly each other.

705: 

706: In 1972 or subsequent years,

707: the observed decrease in fluctuations with country landings,

708: Eq.(\ref{eqn:scaling-growth-rate}),

709: is considerably slower (i.e. $\beta\approx 0.15$) than in the case~(ii).

710: Our empirical results suggest an important consequence for dynamics of stock

711: exploitation:

712: although countries with large fisheries tend to diversify into a wider

713: range of fishing activities leading to smaller relative fluctuations in

714: country landings,

715: fisheries catches reported are more variable than what would be expected

716: if the number of fishers would increase linearly with the country

717: landings---which would correspond to $\beta =0.5$.

718: 

719: 

720: \subsection{Scaling relation between exponents}

721: The country landings does not scale in a simple linear fashion with

722: increasing number of fishers, but instead

723: the typical number of fishers scales with landings as

724: Eq.(\ref{eqn:typical-number-fishers}).

725: The total capture production is broken down by numbers of landing sites

726: (i.e. fishing ports) in a country.

727: The fishery products in landing sites are landed by numbers of local

728: fishers.

729: These production equals the marketing volume of fish at local landing

730: sites.

731: Let us analyze fishery statistics of Japan on the marketing of fishery

732: products in major landing areas \citep{MAFF00-02}.

733: It is virtually apparent in Fig.\ref{fig:9},

734: showing the probability distribution of catches landed at

735: 203 ports~({\sf A}) and the probability density of annual growth rate of

736: port landings~({\sf B}),

737: that the same behavior holds for subsystems, i.e. the single country's

738: fisheries sector, as the world capture fisheries sector.

739: The subsystems or hierarchical systems of capture fisheries sector are

740: expected to be characterized by the same fundamental laws.

741: 

742: To account for the increased variability for highly producing countries

743: (i.e. $\beta\approx 0.15$ for years 1972--2002),

744: we suppose that

745: in a country with $x$ tonnes domestic fisheries production,

746: the width of the annual per capita growth-rate distribution,

747: denoted by $\sigma_0(x)$,

748: scales with landings $x$ as

749: %

750: \begin{equation}

751:  \sigma_0(x)

752:   \propto

753:   x^{\gamma}

754:   \label{eqn:fluctuation-gamma}

755: \end{equation}

756: with an exponent $\gamma$.

757: Motivated by scaling behavior in the exploitation dynamics,

758: I expect that

759: there exist scaling relations among three exponents $\alpha$, $\beta$

760: and $\gamma$ \citep{Widom65a,Widom65b}.

761: Assume that the per capita growth rate is independent (uncorrelated),

762: identically distributed,

763: though individual fishers may not be right fishing units that produce

764: catches independently of one another;

765: e.g. there are $132$ thousand fishery establishments

766: while $238$ thousand fishers exist in Japan in 2003 \citep{MAFF05}.

767: The central limit theorem is then applied to give

768: \begin{equation}

769:  \sigma_r (x)

770:   =

771:   \frac{\sigma_0(x)}{\sqrt{\overline{n}(x)}},

772: \end{equation}

773: leading to a scaling relation

774: \begin{equation}

775:  \beta

776:   =

777:   \frac{\alpha}2 -\gamma.

778:   \label{eqn:scaling-relationship-beta}

779: \end{equation}

780: For $\alpha = 0.71\pm 0.01$ and

781: $\beta = 0.15\pm 0.02$,

782: Eq.(\ref{eqn:scaling-relationship-beta}) predicts

783: $\gamma = 0.21\pm 0.03$.

784: It is remarkable that,

785: contrary to country landings (Fig.\ref{fig:4}{\sf A}),

786: the per capita landings of highly producing countries fluctuate more

787: than do those of low producing countries

788: [Eq.(\ref{eqn:fluctuation-gamma}) with $\gamma > 0$]:

789: fishers in a country with larger landings risk more than do fishers in a

790: country with smaller landings in capture fisheries.

791: 

792: 

793: \section{Discussion}

794: Studying the biological properties of the FAO global fisheries

795: statistics on landed catches has given ecological insight to the state

796: of world fish stocks: e.g.

797: in Refs \citep{Pauly-etal98,Caddy-Garibaldy00,Pauly-etal05},

798: global conclusions drawn state that a general decline in mean trophic

799: level of marine and inland landings has occurred in many regions,

800: reflecting a change in catch composition from long-lived, high trophic

801: level, piscivorous bottom fishes toward short-lived, low trophic level

802: invertebrates and planktivorous pelagic fishes.

803: In this paper, from the socio-ecological viewpoint,

804: I have analyzed the FAO data and

805: found the universal scaling behavior in the dynamics of stock

806: exploitation.

807: Moreover, we empirically find

808: from Fig.\ref{fig:6} and

809: Eq.(\ref{eqn:typical-number-fishers}),

810: that

811: the per capita production of countries with landings $x$ scales as

812: \begin{equation}

813:  \frac x{\overline{n}(x)}

814:   \propto

815:   x^{1-\alpha},

816: \end{equation}

817: showing the increased per capita catches for highly producing countries

818: (i.e. $1-\alpha > 0$).

819: However, fishing business may have a high-risk premium:

820: the larger their expected revenues in highly producing countries,

821: the larger the fluctuations in annual per capita revenues,

822: which is predicted by Eq.(\ref{eqn:fluctuation-gamma}) with $\gamma > 0$;

823: whereas the empirical data demonstrate

824: decreasing fluctuations in annual domestically-landed catches with

825: increased country landings

826: [Fig.\ref{fig:4}{\sf A} and

827: Eq.(\ref{eqn:scaling-growth-rate}) with $\beta > 0$].

828: The expected fishing risk (i.e. fluctuations in the individual catches)

829: in a country with $x$ tonnes annual domestic landings scales with the

830: individual expected catches:

831: $\sigma_0(x) \propto [x/\overline{n}(x)]^{\delta}$,

832: with an exponent $\delta = 0.7\pm 0.1$,

833: which seems to agree with the value of the exponent $\alpha$.

834: 

835: Though the growth era of the capture fisheries production has been ended

836: during the last decade,

837: the world fishers have been growing in number

838: (35.8 percent from 1990 to 2002).

839: The fishers population to total world population ratio

840: has been increasing for the past

841: thirty years and stood at 60.7 fishers per one thousand people in 2002.

842: In most developing countries of low and middle-income,

843: the number of people employed in the fisheries sector has been

844: growing steadily.

845: In industrialized economies offering occupational alternatives,

846: the numbers of fishers have been on a declining trend or at best

847: stationary.

848: For instance in two important fishing countries, Japan and Norway,

849: fishers have decreased in number by 34.3 and 19.7 percent, respectively,

850: during the 1990 to 2002 period.

851: The increasing trend of world fishers population

852: may suggest that

853: wages received by fishers (mostly in developing countries)

854: exceed

855: the level of wages that fishers could expect to earn in alternative

856: employment opportunities,

857: as commonly understood in economic analysis from the static viewpoint of

858: the common-property fishery.

859: Because fishers would be making money more than their opportunity costs

860: (i.e. resource rent is positive),

861: additional fishers would be attracted to the fishery,

862: until

863: it reaches the ``bionomic'' equilibrium \citep{Gordon54}

864: where total sustained revenue equals total sustained cost.

865: Contrary, if net revenues from fishing are negative, fishers leave the

866: fishery.

867: 

868: The exploitation dynamics of fish stocks exhibits two-phase behavior,

869: which may be according to stock status together with management policy.

870: There was a change in scaling behavior of the standard deviation of

871: production growth rates at 1972.

872: It may be more than coincidence that

873: the emergence of the United Nations' Convention on the Law of the Sea,

874: in the late 1970s,

875: which enabled countries to claim exclusive economic zones reaching 200

876: nautical miles into the open sea,

877: put the responsibility for fisheries resource management squarely with

878: maritime countries,

879: thus ending many decades of fighting over traditional fishing grounds

880: \citep{Johnstone77}.

881: 

882: Another possible cause for the observed transition at 1972 in scaling

883: structure is that,

884: having depleted large valuable stocks,

885: fishing has redirected some effort and added a lot of it on other

886: species lower down the food web.

887: The strategy was advocated in the 1970s to increase fisheries

888: production

889: \citep{Sprague-Arnold72,Garcia-Grainger05}.

890: The phenomenon of change in catch composition, i.e.

891: ``fishing down marine food webs'' \citep{Pauly-etal98},

892: contrasts strikingly with D'Aancona's observation on Mediterranean

893: fisheries \citep{D'Ancona26} that

894: the cessation of the fishery during World War I from 1914--1918 changed

895: the species composition in favor of fish-eating fishes

896: (the frequency of predators decreased with an increase in fishing after

897: the war).

898: The decline in mean trophic level of the world fish stocks leads to

899: increasing variability in catches \citep{Pauly-etal02}.

900: Indeed, we find from the analysis of the FAO data for all countries

901: (excluding China) that

902: the country landings fluctuate more for the periods 1972--2002

903: than do those for the periods 1950--1971.

904: The standard deviation of growth rate of the annual domestically-landed

905: catches for 30 one-year periods between 1972 and 2002

906: shows an increase of more than 10 percent over that for 21 one-year

907: periods between 1950 and 1971:

908: \begin{equation}

909: \overline{\sigma}_{r,\mbox{\tiny 1950--1971}} = 0.138\pm 0.004,\quad

910: \overline{\sigma}_{r,\mbox{\tiny 1972--2002}} = 0.156\pm 0.003,

911: \end{equation}

912: whereas the average yearly growth rate declined significantly after

913: 1972:

914: \begin{equation}

915: \overline{r}_{\mbox{\tiny 1950--1971}} = 0.077,\quad

916: \overline{r}_{\mbox{\tiny 1972--2002}} = 0.026.

917: \end{equation}

918: Besides highly variable recruitment of planktivorous pelagic fishes that

919: is strongly influenced by environmental fluctuations,

920: the schooling behavior common among such pelagics is a possible cause of

921: variation in catch in purse seine fisheries.

922: Some fishers catch more than others:

923: the distribution of fishers' yearly total landed value of catch in a

924: commercial purse seine fishery is skewed to the right \citep{Hilborn85}.

925: School-size distributions of pelagic fishes, i.e. catch-per-haul

926: distributions in purse seine fisheries, are generally heavy-tailed,

927: and a typical school-size

928: (rigorously defined as a crossover size from power-law to exponential

929: decay of the school-size distribution)

930: is proportional to the abundance of the fish

931: stock \citep{Niwa03,Niwa04}, seeming to be consistent with

932: Eq.(\ref{eqn:fluctuation-gamma}) with $\gamma > 0$.

933: 

934: The global potential harvest (e.g. maximum sustainable yield) may be one

935: of the causes for the observed transition in scaling behavior of fishery

936: landings.

937: \citet{Grainger-Garcia96} estimated the most likely potential of

938: conventional marine species to range from 80 to 100 million tonnes,

939: which has indeed been reached probably in the 1970s and is unlikely to

940: change in the next 20--30 years \citep{Garcia-Grainger05}.

941: Undeveloped resource fisheries, producing much less than their

942: potential, decreased rapidly to zero by the middle of the 1970s

943: \citep{Grainger-Garcia96}.

944: The latest analysis of the state of resources \citep{Garcia-etal05}

945: indicates that, in 2003, approximately half the world's stocks are

946: exploited at or close to their maximum, and about 25 percent of them are

947: exploited either below or above such maximum.

948: 

949: The same scaling laws as found here describe a number of other complex

950: organizations \citep{Stanley02}, such as

951: business firms \citep{Stanley96},

952: gross domestic product \citep{Lee98},

953: university research budgets \citep{Plerou99},

954: and

955: bird populations \citep{Keitt-etal02},

956: suggesting the underlying universal mechanism.

957: It is no wonder that commercial fisheries world-wide resemble

958: competitive economic activities.

959: In the same way that market forces operate in business, 

960: the stock exploitation appears to keep competition among fishers

961: strong enough to leave capture fisheries with a shrinking resource base.

962: According to FAO press release \citep{double},

963: most small-scale fishers find it increasingly difficult to survive in an

964: over-exploited environment.

965: 

966: \citet{Amaral98} proposed a model for the growth of organizations with

967: complex internal structure.

968: In the model, each organization

969: (e.g. firm, university, or bird population)

970: is made up of units.

971: The units composing the system have a complex evolving structure,

972: e.g.

973: business firms competing in an economy are comprised of divisions;

974: universities are composed of schools or colleges;

975: population of breeding birds living in a given ecosystem is composed

976: of groups living in different areas.

977: Their model postulates that

978: the units grow through an independent, Gaussian-distributed, random

979: multiplicative process.

980: Later this postulate empirically proved not to be accurate

981: \citep{Matia-etal04}.

982: Thus the model proposed by \citet{Amaral98} can now be regarded

983: as a first step towards the explanation of the growth dynamics of

984: organizations.

985: Scaling concepts offer an avenue to study heterogeneous assemblies in

986: ecological and economic systems for which the microscopic processes are

987: (probably) not knowable, except in terms of their statistical properties

988: \citep{Chave-Levin03}.

989: 

990: 

991: \newpage

992: %REFERENCES

993: \begin{thebibliography}{99}

994: %

995:  \bibitem[Amaral et al.(1998)]{Amaral98}

996: 	 Amaral, L. A. N., Buldyrev, S. V., Havlin, S., Salinger, M. A.

997: 	 and Stanley, H. E., 1998.

998: 	 Power law scaling for a system of interacting units with

999: 	 complex internal structure.

1000: 	 {Phys. Rev. Lett.} {80}, 1385--1388.

1001: %

1002:  \bibitem[Axtell(2001)]{Axtell01}

1003: 	 Axtell, R. L., 2001.

1004: 	 Zipf distribution of U.S. firm.

1005: 	 {Science} {293}, 1818--1820.

1006: %

1007:  \bibitem[Beddington and Kirkwood(2005)]{Beddington-Kirkwood05}

1008: 	 Beddington, J. and Kirkwood, G. (Editors), 2005.

1009: 	 {Fisheries: a Future?}

1010: 	 {Phil. Trans. R. Soc. B} {360} (1453).

1011: %

1012:  \bibitem[Bj{\o}rnstad and Grenfell(2001)]{Bjornstad-Grenfell01}

1013:          Bj{\o}rnstad, O. N. and Grenfell, B. T., 2001.

1014:          Noisy clockwork: time series analysis of population

1015: 	 fluctuations in animals.

1016:          {Science} {293}, 638--643.

1017: %

1018:  \bibitem[Bj{\o}rnstad et al.(1999)]{Bjornstad-etal99}

1019:          Bj{\o}rnstad, O. N., Fromentin, J.-M., Stenseth, N. C. and

1020: 	 Gj{\o}s{\ae}ter, J., 1999.

1021:          Cycles and trends in cod populations.

1022:          {Proc. Natl. Acad. Sci. USA} {96}, 5066--5071.

1023: %

1024:  \bibitem[Botsford et al.(1997)]{Botsford-etal97}

1025: 	 Botsford, L. W., Castilla, J. C. and Peterson, C. H., 1997.

1026: 	 The management of fisheries and marine ecosystems.

1027: 	 {Science} {277}, 509--515.

1028: %

1029:  \bibitem[Buldyrev et al.(1997)]{buldyrev-etal97}

1030:          Buldyrev, S. V., Amaral, L. A. N., Havlin, S., Leschhorn, H.,

1031:          Maass, P., Salinger, M. A., Stanley, H. E. and Stanley, M. H. R.,

1032:          1997.

1033:          Scaling behavior in economics: II. modeling of company growth.

1034:          {J. Phys. I France} {7}, 635--650.

1035: %

1036:  \bibitem[Caddy and Garibaldy(2000)]{Caddy-Garibaldy00}

1037: 	 Caddy, J. F. and Garibaldy, L., 2000.

1038: 	 Apparent changes in the trophic composition of world marine

1039: 	 harvests: the perspective from the FAO capture database.

1040: 	 {Ocean and Coastal Management} {43}, 615--655.

1041: %

1042:  \bibitem[Chave and Levin(2003)]{Chave-Levin03}

1043: 	 Chave, J. and Levin, S., 2003.

1044: 	 Scale and scaling in ecological and economic systems.

1045: 	 {Environ. Resource Econ.} {26}, 527--557.

1046: %

1047:  \bibitem[Conklin and Kolberg(1994)]{Conklin-Kolberg94}

1048:          Conklin, J. E. and Kolberg, W. C., 1994.

1049:          Chaos for the halibut?

1050:          {Mar. Resour. Econ.} {9}, 153--182.

1051: %

1052:  \bibitem[D'Ancona(1926)]{D'Ancona26}

1053: 	 D'Ancona, U., 1926.

1054: 	 Dell'influenza della stasi peschereccia del periodo 1914-1918

1055: 	 sul patrimonio ittico dell'Alto Adriatico.

1056: 	 {Regio Comitato Talassografico Italiano, Memoria} {126}.

1057: %

1058:  \bibitem[D'Ancona(1954)]{D'Ancona54}

1059: 	 D'Ancona, U., 1954.

1060: 	 {The Struggle for Existence}. E. J. Brill, Leiden.

1061: %

1062:  \bibitem[Dixon et al.(1999)]{Dixon-etal99}

1063:          Dixon, P. A., Milicich, M. J. and Sugihara, G., 1999.

1064:          Episodic fluctuations in larval supply.

1065:          {Science} {283}, 1528--1530.

1066: %

1067:  \bibitem[Ehrhardt(2005)]{Ehrhardt05}

1068: 	 Ehrhardt, N. M., 2005.

1069: 	 Introduction to the world conference on the ``scientific and

1070: 	 technical bases for the sustainability of fisheries''.

1071: 	 {Bull. Mar. Sci.} {76}, 167--170.

1072: %

1073:  \bibitem[FAO(1998)]{double}

1074: 	 FAO, 1998.

1075: 	 Number of fishers doubled since 1970.

1076: 	 {FAO Press Release 98/09}.

1077: 	 web-site at

1078: 	 http://www.fao.org/WAICENT/OIS/PRESS{\_}NE/ PRESSENG/1998/pren9809.htm.

1079: %

1080:  \bibitem[FAO(1999)]{NoFishers}

1081: 	 FAO, 1999.

1082: 	 Numbers of fishers, 1970--1996.

1083: 	 {FAO Fisheries Circular} No.929, Rev.1.

1084: 	 FAO, Rome.

1085: 	 I here use the CSV (comma separated value) data of World

1086: 	 Resources Institute, at Internet site

1087: 	 ``{EarthTrends/Fisheries: People actively fishing, number}''

1088: 	 (http://earthtrends.wri.org/text/COA/variables/ 698.csv).

1089: %

1090:  \bibitem[FAO(2004a)]{FaoFishStat+}

1091: 	 FAO, 2004a.

1092: 	 {Fishstat Plus:

1093: 	 Universal software for fishery statistical time series}

1094: 	 (version 2.3)

1095: 	 with the dataset ``{Capture production 1950--2002}''.

1096: 	 web-site at

1097: 	 http://www.fao.org/fi/statist/fisoft/FISHPLUS.asp.

1098: 	 Data included in the dataset are published as

1099: 	 ``{FAO Yearbook of Fishery Statistics, Capture production

1100: 	 2002}, Vol.94/1 (2004).'' FAO, Rome.

1101: %

1102:  \bibitem[FAO(2004b)]{SOFIA02}

1103: 	 FAO, 2004b.

1104: 	 {The State of World Fisheries and Aquaculture 2004}.

1105: 	 web-site at

1106:          http://www.fao.org/sof/sofia/.

1107: %

1108:  \bibitem[Frisch and Sornette(1997)]{Frisch-Sornette97}

1109: 	 Frisch, U. and Sornette, D., 1997.

1110: 	 Extreme deviations and applications.

1111: 	 {J. Phys. I France} {7}, 1155--1171.

1112: %

1113:  \bibitem[Garcia and Grainger(2005)]{Garcia-Grainger05}

1114: 	 Garcia, S. M. and Grainger, R. J. R., 2005.

1115: 	 Gloom and doom? the future of marine capture fisheries.

1116: 	 {Phil. Trans. R. Soc. B} {360}, 21--46.

1117: %

1118:  \bibitem[Garcia et al.(2005)]{Garcia-etal05}

1119: 	 Garcia, S. M., de Leiva Moreno, I. and Grainger, R. J. R., 2005.

1120: 	 Global trends in the state of marine fisheries resources

1121: 	 1974--2003.

1122: 	 In: {FAO}, {Review of the State of World Marine Fishery

1123: 	 Resources}.

1124: 	 {FAO Fish. Tech. Pap.} {457}, 10--14.

1125: %

1126:  \bibitem[Gibrat(1931)]{Gibrat31}

1127: 	 Gibrat, R., 1931.

1128: 	 {Les In\'egaliti\'es \'Economiques;

1129: 	 Applications:

1130: 	 aux in\'egaliti\'es des richesses, \`a la concentration des

1131: 	 extreprises, aux populations des villes, aux statistiques des

1132: 	 familles, etc., d'une loi nouvelle, la loi de l'effet

1133: 	 proportionnel}.

1134: 	 Librarie du Recueil Sirey, Paris.

1135: %

1136:  \bibitem[Gordon(1954)]{Gordon54}

1137: 	 Gordon, H. S. 1954.

1138: 	 The economics of a common property resource: the fishery.

1139: 	 {J. Polit. Econ.} {62}, 124--142.

1140: %

1141:  \bibitem[Grainger and Garcia(1996)]{Grainger-Garcia96}

1142: 	 Grainger, R. J. R. and Garcia, S. M., 1996.

1143: 	 Chronicles of marine fishery landings (1950--1994): trend

1144: 	 analysis and fisheries potential.

1145: 	 {FAO Fish. Tech. Pap.} {359}.

1146: %

1147:  \bibitem[Higgins et al.(1997)]{Higgins-etal97}

1148:          Higgins, K., Hastings, A., Sarvela, J. N. and Botsford, L. W.,

1149: 	 1997.

1150:          Stochastic dynamics and deterministic skeletons: population

1151: 	 behavior of Dungeness crab.

1152:          {Science} {276}, 1431--1435.

1153: %

1154:  \bibitem[Hilborn(1985)]{Hilborn85}

1155: 	 Hilborn, R., 1985.

1156: 	 Fleet dynamics and individual variation: why some people catch

1157: 	 more fish than others.

1158: 	 {Can. J. Fish. Aquat. Sci.} {42}, 2--13.

1159: %

1160:  \bibitem[Hsieh et al.(2005)]{Hsieh-etal05}

1161: 	 Hsieh, C., Glaser, S. M., Lucas, A. J. and Sugihara, G., 2005.

1162: 	 Distinguishing random environmental fluctuations from

1163: 	 ecological catastrophes for the North Pacific Ocean.

1164: 	 {Nature} {435}, 336--340.

1165: %

1166:  \bibitem[Johnstone(1977)]{Johnstone77}

1167: 	 Johnstone, K., 1977.

1168: 	 {The Aquatic Explorers: A History of the Fisheries Research

1169: 	  Board of Canada}.

1170: 	 University of Toronto Press, Toronto.

1171: %

1172:  \bibitem[Keitt and Stanley(1998)]{Keitt-Stanley98}

1173: 	 Keitt, T. and Stanley, H. E., 1998.

1174: 	 Dynamics of North American breeding bird populations.

1175: 	 {Nature} {393}, 257--260.

1176: %

1177:  \bibitem[Keitt {et al.}(2002)]{Keitt-etal02}

1178: 	 Keitt, T. H., Amaral, L. A. N., Buldyrev, S. V. and Stanley, H. E.,

1179: 	 2002.

1180: 	 Scaling in the growth of geographically subdivided populations:

1181: 	 invariant patterns from a continent-wide biological survey.

1182: 	 {Phil. Trans. R. Soc. Lond. B} {357}, 627--633.

1183: %

1184:  \bibitem[Laherr{\` e}re and Sornette(1998)]{Laherrere-Sornette98}

1185: 	 Laherr{\` e}re J. H. and Sornette, D., 1998.

1186: 	 Stretched exponential distributions in nature and economy:

1187: 	 ``fat tails'' with characteristic scales.

1188: 	 {Eur. Phys. J. B} {2}, 525--539.

1189: %

1190:  \bibitem[Lee et al.(1998)]{Lee98}

1191: 	 Lee, Y., Amaral, L. A. N., Canning, D., Meyer, M. and

1192: 	 Stanley, H. E., 1998.

1193: 	 Universal features in the growth dynamics of complex

1194: 	 organizations.

1195: 	 {Phys. Rev. Lett.} {81}, 3275--3278.

1196: %

1197:  \bibitem[Lotka(1925)]{Lotka25}

1198:          Lotka, A. J., 1925.

1199: 	{Elements of Physical Biology}.

1200:          Williams \& Wilkins Co., Baltimore.

1201: %

1202:  \bibitem[Matia et al.(2004)]{Matia-etal04}

1203: 	 Matia, K., Fu, D., Buldyrev, S. V., Pammolli, F.,

1204: 	 Riccaboni, M. and Stanley, H. E.,

1205: 	 2004.

1206: 	 Statistical properties of business firms structure and growth.

1207: 	 {Europhys. Lett.} {67}, 498--503.

1208: %

1209:  \bibitem[Matsuda et al.(1991)]{Matsuda-etal91}

1210: 	 Matsuda. H., Wada, T., Takeuchi, Y. and Matsumiya, Y., 1991.

1211:          Alternative models for species replacement of pelagic fishes.

1212:          {Res. Popul. Ecol.} {33}, 41--56.

1213: %

1214:  \bibitem[Matsuda et al.(1992)]{Matsuda-etal92}

1215:          Matsuda, H., Wada, T., Takeuchi, Y. and Matsumiya, Y., 1992.

1216:          Model analysis of the effect of environmental fluctuation on

1217: 	 the species replacement pattern of pelagic fishes under

1218: 	 interspecific competition.

1219:          {Res. Popul. Ecol.} {34}, 309--319.

1220: %

1221:  \bibitem[May(1974)]{May74}

1222:          May, R. M., 1974.

1223:          Biological populations with nonoverlapping generations: stable

1224:          points, stable cycles and chaos.

1225:          {Science} {186}, 645--647.

1226: %

1227:  \bibitem[May(1976)]{May76}

1228:          May, R. M., 1976.

1229:          Simple mathematical models with very complicated dynamics.

1230:          {Nature} {261}, 459--467.

1231: %

1232:  \bibitem[McQuaid et al.(1996)]{McQuaid-etal96}

1233:          McQuaid, J., Marshall, B., Schleifstein, M. and Jackson, T., 1996.

1234:          {Oceans of Trouble}.

1235:          The Times-Picayune, New Orleans.

1236:          March 24--31, 1996.

1237:          available at

1238:          Internet site ``{1997 Pulitzer Prizes}''

1239:          (http://www.pulitzer.org/year/1997/ public-service/works/).

1240: %

1241:  \bibitem[Ministry of Agriculture Forestry and Fisheries of Japan(2002,\ 2003,\ 2004)]{MAFF00-02}

1242:          Ministry of Agriculture Forestry and Fisheries of Japan,

1243:          2002, 2003 and 2004.

1244:          {Yearbook of Statistics on the Marketing of Fishery Products

1245:          (2000, 2001 and 2002)}.

1246:          Association of Agriculture-Forestry Statistics, Tokyo.

1247:          Data downloadable at:

1248:          http://www.tdb.maff.go.jp/toukei/a02smenu?TouID=C005.

1249: 	 (in Japanese)

1250: %

1251:  \bibitem[Ministry of Agriculture Forestry and Fisheries of Japan(2005)]{MAFF05}

1252:          Ministry of Agriculture Forestry and Fisheries of Japan,

1253:          2005.

1254:          {The 11th Fishery Census of Japan 2003}.

1255:          Association of Agriculture-Forestry Statistics, Tokyo.

1256:          Summary of survey results available at:

1257:          http://www.maff.go.jp/ toukei/sokuhou/data/gyogyou-census2003-v/gyogyou-census2003-v.htm.

1258: 	 (in Japanese)

1259: %

1260:  \bibitem[Niwa(2003)]{Niwa03}

1261: 	 Niwa, H.-S., 2003.

1262: 	 Power-law versus exponential distributions of animal group

1263: 	 sizes.

1264: 	 {J. Theor. Biol.} {224}, 451--457.

1265: %

1266:  \bibitem[Niwa(2004)]{Niwa04}

1267: 	 Niwa, H.-S., 2004.

1268: 	 Space-irrelevant scaling law for fish school sizes.

1269: 	 {J. Theor. Biol.} {228}, 347--357.

1270: %

1271:  \bibitem[Pauly et al.(1998)]{Pauly-etal98}

1272: 	 Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. and

1273: 	 Torres, Jr., F., 1998.

1274: 	 Fishing down marine food webs.

1275: 	 {Science} {279}, 860--863.

1276: %

1277:  \bibitem[Pauly et al.(2002)]{Pauly-etal02}

1278: 	 Pauly, D., Christensen, V., Gu\'enette, S., Pitcher, T. J.,

1279: 	 Sumaila, U. R., Walters, C. J., Watson, R. and Zeller, D., 2002.

1280: 	 Towards sustainability in world fisheries.

1281: 	 {Nature} {418}, 689--695.

1282: %

1283:  \bibitem[Pauly et al.(2005)]{Pauly-etal05}

1284: 	 Pauly, D., Watson, R. and Alder, J., 2005.

1285: 	 Global trends in world fisheries: impacts on marine ecosystems

1286: 	 and food security.

1287: 	 {Phil. Trans. R. Soc. B} {360}, 5--12.

1288: %

1289:  \bibitem[Plerou et al.(1999)]{Plerou99}

1290: 	 Plerou, V., Amaral, L. A. N., Gopikrishnan, P., Meyer, M. and

1291: 	 Stanley, H. E., 1999.

1292: 	 Similarities between the growth dynamics of university research

1293: 	 and of competitive economic activities.

1294: 	 {Nature} {400}, 433--437.

1295: %

1296:  \bibitem[S{\ae}ther et al.(2000)]{Saether-etal00}

1297:          S{\ae}ther, B.-E., Tufto, J., Engen, S., Jerstad, K.,

1298: 	 R{\o}stad, O. W. and Sk{\aa}tan, J. E., 2000.

1299:          Population dynamical consequences of climate change for a small

1300: 	 temperate songbird.

1301:          {Science} {287}, 854--856.

1302: %

1303:  \bibitem[Smith(1994)]{Smith94}

1304: 	 Smith, T. D., 1994.

1305: 	 {Scaling Fisheries: the science of measuring the effects of

1306:          fishing, 1855-1955}.

1307: 	 Cambridge University Press, Cambridge.

1308: %

1309: \bibitem[Sprague and Arnold(1972)]{Sprague-Arnold72}

1310: 	Sprague, L. M. and Arnold, J. H., 1972.

1311: 	Trends in the use and prospects for the future harvest of world

1312: 	fisheries resources.

1313: 	{J. Am. Oil Chem. Soc.} {49}, 345--350.

1314: %

1315: \bibitem[Stanley et al.(2002)]{Stanley02}

1316: 	 Stanley, H. E., Amaral, L. A. N., Buldyrev, S. V.,

1317: 	 Gopikrishnan, P., Plerou, V. and Salinger, M. A., 2002.

1318:          Self-organized complexity in economics and finance.

1319: 	 {Proc. Natl. Acad. Sci. USA} {99}, suppl.~1,

1320: 	 2561--2565.

1321: %

1322: \bibitem[Stanley et al.(1996)]{Stanley96}

1323: 	 Stanley, M. H. R., Amaral, L. A. N., Buldyrev, S. V., Havlin,

1324: 	 S., Leschhorn, H., Maass, P., Salinger, M. A. and Stanley, H. E.,

1325: 	 1996.

1326: 	 Scaling behaviour in the growth of companies.

1327: 	 {Nature} {379}, 804--806.

1328: %

1329: \bibitem[Symes(2005)]{Symes05}

1330: 	Symes, D., 2005.

1331: 	Fishing, the environment and the media.

1332: 	{Fish. Res.} {73}, 13--19.

1333: %

1334: \bibitem[Takeuchi et al.(1992)]{Takeuchi-etal92}

1335:         Takeuchi, Y., Oshime, Y. and Matsuda, H., 1992.

1336:         Persistence and periodic orbits of a three-competitor model with

1337:         refuges.

1338:         {Math. Biosci.} {108}, 105--125.

1339: %

1340: \bibitem[Turchin(2003)]{Turchin03}

1341:         Turchin, P., 2003.

1342:         {Complex Population Dynamics:

1343:         a theoretical/empirical synthesis}.

1344:         Princeton University Press, Princeton.

1345: %

1346:  \bibitem[United Nations Population Division(2004)]{unpp04}

1347: 	 United Nations Population Division of the Department of

1348: 	 Economic and Social Affairs, 2004.

1349: 	 {World Population Prospects: the 2004 revision}.

1350: 	 web-site at http://esa.un.org/unpp.

1351: %

1352:  \bibitem[Volterra(1926)]{Volterra26}

1353: 	 Volterra, V., 1926.

1354: 	 Fluctuations in the abundance of a species considered

1355: 	 mathematically.

1356: 	 {Nature} {118}, 558--560.

1357: %

1358:  \bibitem[Watson and Pauly(2001)]{WatsonPauly01}

1359: 	 Watson, R. and Pauly, D., 2001.

1360: 	 Systematic distortions in world fisheries catch trends.

1361: 	 {Nature} {414}, 534--536.

1362: %

1363:  \bibitem[Weibull(1951)]{Weibull51}

1364: 	 Weibull, W., 1951.

1365: 	 A statistical distribution function of wide applicability.

1366: 	 {J. Appl. Mech.} {18}, 293--297.

1367: %

1368:  \bibitem[Weisbuch et al.(1997)]{Weisbuch-etal97}

1369: 	 Weisbuch, G., Stanley, E. A., Duchateau-Nguyen, G., Antona, M.

1370: 	 and Cl{\'e}ment-Pitiot, H.,

1371: 	 1997.

1372: 	 Influence of capital inertia on renewable resource depletion.

1373: 	 {Theory Biosci.} {116}, 97--117.

1374: %

1375:  \bibitem[Widom(1965a)]{Widom65a}

1376: 	 Widom, B., 1965a.

1377: 	 Surface tension and molecular correlations near the critical

1378: 	 point.

1379: 	{J. Chem. Phys.} {43}, 3892--3897.

1380: %

1381:  \bibitem[Widom(1965b)]{Widom65b}

1382: 	 Widom, B., 1965b.

1383: 	 Equation of state in the neighborhood of the critical point.

1384: 	 {J. Chem. Phys.} {43}, 3898-3905. 

1385: %

1386:  \bibitem[Wilson et al.(1991)]{Wilson-etal91}

1387:          Wilson, J. A., J. French, P. Kleban, S. R. McKay and Townsend, R.,

1388: 	 1991.

1389:          Chaotic dynamics in multiple species fishery: a model of

1390: 	 community predation.

1391: 	 {Ecol. Model.} {58}, 303--322.

1392: %

1393:  \bibitem[Zipf(1949)]{Zipf47}

1394:          Zipf, G. K., 1949.

1395:          {Human Behavior and the Principle of Least Effort}.

1396:          Addison-Wesley, Reading, Massachusetts.

1397: %

1398: \end{thebibliography}

1399: %

1400: 

1401: 

1402: \newpage

1403: \begin{flushleft}

1404: {\bf Figure legends}

1405: \end{flushleft}

1406: 

1407: \begin{itemize}

1408:  \setlength{\itemsep}{10pt}

1409:                  %

1410: %

1411: \item[Fig.1]

1412:  World fisheries trend including ($\circ$)/excluding ($\bullet$) China.

1413: %

1414:  ({\sf A})~Semi-logarithmic scale plot of global capture trend.

1415:  Exponentially growing behavior with two growth-rates is observed.

1416:  There is a knee in the curve at 1972,

1417:  probably figuring the boundary between two distinct regimes of the

1418:  bio-economies.

1419:  The solid lines are fits of an exponential function

1420: [$\propto \exp (Rt)$]

1421:  to the data~($\bullet$) with average yearly growth rates

1422:  $R = 0.059\pm 0.001$ and $0.024\pm 0.001$

1423:  for the periods 1950--1971 and 1972--1989, respectively.

1424:  A plateau in the 1990s follows immediately

1425:  (to be precise $R = -0.002\pm 0.003$).

1426:  Fitting the data~($\circ$) gives

1427:  $R = 0.058\pm 0.001$ and $0.017\pm 0.001$

1428:  for the periods 1950--1971 and 1972--2002, respectively

1429:  (depicted by broken lines).

1430: %

1431:  ({\sf B})~Trend of the world fisher population $N$.

1432:  The ordinate represents $\log_{10} N$.

1433:  The broken and solid lines depict exponential fits to the data

1434:  including and excluding China,

1435:  giving population growth rate

1436:  $R_N = 0.035\pm 0.002$ and

1437:  $R_N = 0.028\pm 0.001$

1438:  for the period 1970--2002, respectively

1439:  (annual growth rate of the fisher population world-wide is defined as

1440:  same as the global harvest case).

1441:  The inserted figure shows the fisher-to-population ratio.

1442:  The earth's human population data from 1970 to 2000 were retrieved from

1443:  the web-site of \citet{unpp04}.

1444:  The human population has been growing at an average rate

1445:  of 1.67 percent per year (excluding China, 1.74 percent)

1446:  since 1970 to 2000.

1447: %

1448: \item[Fig.2]

1449:  Fisheries production and employment statistics.

1450:  Data exclude China's statistics.

1451: %

1452:  ({\sf A})~Semi-logarithmic scale plot of probability distribution of

1453:  country landings,

1454:  $P(x)$.

1455:  The abscissa is chosen as logarithmically scaled landings

1456:  $\ln (x/\hat{x})/s_x$.

1457:  The data from 6 one-year periods between 1950 and 2000 are plotted,

1458:  where geometric means $\hat{x}$

1459:  of landings (in thousand tonnes) and geometric standard deviations $s_x$

1460:  are as follows:

1461:  $\hat{x} = 6.0$,

1462:  9.8,

1463:  15.4,

1464:  16.9,

1465:  20.1 and

1466:  20.4, and

1467:  $s_x = 2.5$,

1468:  2.6,

1469:  2.7,

1470:  2.9,

1471:  2.9 and

1472:  2.9,

1473:  from years

1474:  1950~$(\square)$,

1475:  1960~$(\lozenge)$,

1476:  1970~$(\vartriangle)$,

1477:  1980~$(\blacksquare)$,

1478:  1990~$(\blacklozenge)$ and

1479:  2000~$(\blacktriangle)$, respectively.

1480:  Log-scaled landings fall onto

1481:  the standardized Gaussian distribution (dashed line)

1482:  with zero mean and unit variance.

1483: %

1484:  ({\sf B})~Semi-logarithmic scale plot of probability distributions of

1485:  the number of fishers and fish farmers in countries,

1486:  $P(n)$.

1487:  The abscissa is chosen as logarithmically scaled fishers population

1488:  $\ln (n/\hat{n})/s_n$,

1489:  where geometric means $\hat{n}$

1490:  of the number of fishers (in thousand fishers) and geometric standard

1491:  deviations $s_n$ are as follows:

1492:  $\hat{n} = 6.7$,

1493:  8.1 and

1494:  10.2, and

1495:  $s_n = 2.1$,

1496:  2.1 and

1497:  2.0,

1498:  from years

1499:  1970~$(\vartriangle)$,

1500:  1980~$(\blacksquare)$ and

1501:  1990~$(\blacklozenge)$, respectively.

1502:  Log-scaled fishers population fall onto

1503:  the standardized Gaussian distribution (dashed line).

1504:  The log-normal distribution holds robustly for fishers as well as landings.

1505: %

1506: \item[Fig.3]

1507:  Exploitation dynamics of fish stocks.

1508: %

1509: ({\sf A})~Semi-logarithmic scale plot of probability density $p(r|x)$ of

1510:  the annual growth rate for eight different bins of

1511:  fisheries production (in tonnes):

1512:  $x < 10^{2.5}\; (\circ)$,

1513:  $10^{2.5} \leq x <10^3\; (\square)$,

1514:  $10^3 \leq x <10^{3.5}\; (\lozenge)$,

1515:  $10^{3.5} \leq x < 10^4\; (\vartriangle)$,

1516:  $10^4 \leq x <10^{4.5}\; (\blacksquare)$,

1517:  $10^{4.5} \leq x < 10^5\; (\blacklozenge)$,

1518:  $10^5 \leq x < 10^{5.5}\; (\blacktriangle)$,

1519:  $10^{5.5} \leq x\; (\bullet)$.

1520:  The data are aggregated over all 21 one-year periods between 1950 and

1521:  1971.

1522:  The distribution decays with fatter tails than for a Gaussian.

1523:  The broken line is a fit of a Laplace distribution

1524:  [Eq.(\ref{eqn:laplace})] for the period, with

1525:  $\overline{r} =0.077$ and $\sigma_r = 0.138\pm 0.004$.

1526:  It is depicted that $p(r|x)$ does not depend on $x$.

1527: %

1528: ({\sf B})~Semi-logarithmic scale plot of probability density of the

1529:  growth rate for countries with small and large landings

1530:  in the years 1972--1989/1990--2002:

1531:  $10^3 \leq x <10^4\; (\lozenge /\vartriangle)$ and

1532:  $10^5 \leq x < 10^6\; (\blacklozenge /\blacktriangle)$ in tonnes.

1533:  The broken and solid lines are Laplacian fits using the means

1534:  $\overline{r} = 0.035$ and $0.016$

1535:  and standard deviations

1536:  $\sigma_r = 0.203\pm 0.004$ and $0.117\pm 0.004$,

1537:  calculated from data for countries with small and large landings in the

1538:  years 1972--1989, respectively.

1539:  Fits with Eq.(\ref{eqn:laplace})

1540:  for small and large landings in 1990--2002

1541:  give the means

1542:  $\overline{r} = 0.034$ and $0.005$

1543:  and standard deviations

1544:  $\sigma_r = 0.219\pm 0.009$ and $0.118\pm 0.007$, 

1545:  overlapping broken and solid lines in the graph.

1546: %

1547: \item[Fig.4]

1548:  Scaling in stock exploitation dynamics.

1549: %

1550: ({\sf A})~Standard deviation of the one-year growth rates of the 21 annual

1551:  intervals from the period 1950--1971~($\circ$), of the 17 annual

1552:  intervals from the period 1972--1989~($\bullet$), and of the 12 annual

1553:  intervals from the period 1990--2002~($\blacksquare$), as a function of

1554:  the landings $x$

1555:  (plotted on a double-logarithmic scale).

1556:  Each data point indicates the width $\sigma_r (x)$ of the growth-rate

1557:  distribution for countries with approximately the same landings

1558:  (binned evenly spaced on a logarithmic scale).

1559:  The long-dashed horizontal line depicts the average of

1560:  the growth fluctuations for years 1950--1971 ($\sigma_r = 0.138$),

1561:  showing $\beta = 0$.

1562:  The solid and broken lines are fits with

1563:  Eq.(\ref{eqn:scaling-growth-rate}) to the data from

1564:  years 1972--1989 ($\beta = 0.14\pm 0.02$) and

1565:  1990--2002 ($\beta = 0.15\pm 0.02$), respectively.

1566:  There are cutoffs at the small landings in power laws.

1567:  Error bars of one standard deviation about each data point are shown.

1568: %

1569: ({\sf B})~Scaled probability density

1570:  $p_{\mbox{\scriptsize sc}}$

1571:  as a function of the scaled growth rate

1572:  $r_{\mbox{\scriptsize sc}}$

1573:  for seven different bins of production in the years 1972--1989 (in

1574:  tonnes):

1575:  $10^3 \leq x <10^{3.5}\; (\square)$,

1576:  $10^{3.5} \leq x < 10^4\; (\lozenge)$,

1577:  $10^4 \leq x <10^{4.5}\; (\vartriangle)$,

1578:  $10^{4.5} \leq x < 10^5\; (\blacksquare)$,

1579:  $10^5 \leq x < 10^{5.5}\; (\blacklozenge)$,

1580:  $10^{5.5} \leq x < 10^6\; (\blacktriangle)$,

1581:  $10^6 \leq x\; (\bullet)$.

1582:  The values are scaled by using the measured values of

1583:  $\overline{r}(x)$ and $\sigma_r (x)$,

1584:  and plotted on a semi-logarithmic scale.

1585:  All the data collapse upon the universal curve

1586:  $p_{\mbox{\scriptsize sc}} = \exp

1587:  \left(-\left|r_{\mbox{\scriptsize sc}}\right|\right)$

1588:  (broken line).

1589:  A data collapse can be obtained for 1990--2002 as well.

1590: %

1591: \item[Fig.5]

1592:  Growth rates of country landings for longer time scales.

1593:  The data points are aggregated over the entire period, 1950--2002,

1594:  for all countries (excluding China).

1595: %

1596:  ({\sf A})~Probability densities $p(r_{\Delta t})$ of $\Delta t$-year

1597:  growth rates are shown

1598:  for $\Delta t = 1$ ($\square$),

1599:  10 ($\lozenge$),

1600:  20 ($\vartriangle$)

1601:  and

1602:  30 years ($\blacksquare$),

1603:  plotted on a semi-logarithmic scale.

1604:  Averages and standard deviations of the $\Delta t$-year growth rate

1605:  are as follows:

1606:  $\overline{r_{\Delta t}} = 0.04$, 0.39, 0.73 and 1.08,

1607:  and 

1608:  $\sigma_{\Delta t} = 0.27$, 0.80, 0.99 and 1.11

1609:  for the same time scales.

1610:  The solid line indicates a Laplace distribution

1611:  [Eq.(\ref{eqn:laplace-entire-period})] with

1612:  $\overline{r_1}$ and $\sigma_1$.

1613:  It is visually apparent that the distributions approach a Gaussian as

1614:  $\Delta t$ increases.

1615: %

1616:  ({\sf B})~Semi-logarithmic scale plot of probability distributions of

1617:  the normalized growth rates $g_{\Delta t}$.

1618:  The broken line shows the standardized Gaussian distribution.

1619: %

1620:  ({\sf C})~Double-logarithmic scale plot of complementary cumulative

1621:  distributions

1622:  $P(\geq g_{\Delta t})$ for the positive tails

1623:  of the probability densities shown in Fig.{\sf B}.

1624:  The complementary CDFs for the negative tails have the same functional

1625:  forms as for the positive tails.

1626: %

1627:  ({\sf D})~Fractional moments for the normalized growth rates for the

1628:  same time scales:

1629:  $\Delta t = 1$ (long-dashed line),

1630:  10 (medium-dashed line),

1631:  20 (short-dashed line)

1632:  and

1633:  30 years (dotted line).

1634:  The solid line shows the Gaussian moments.

1635: %

1636: \item[Fig.6]

1637:  Statistical analysis of the number of fishers.

1638:  Data exclude China's statistics.

1639: %

1640:  ({\sf A})~Number of fishers versus country landings (in tonnes).

1641:  The ordinate and the abscissa are chosen as

1642:  $\log_{10} n$ and $\log_{10} x$, respectively.

1643:  Data are aggregated for seven years (1990--1996).

1644:  The rectangle in gray reads the inter-quantile range of binned data

1645:  and the slit indicates the median,

1646:  showing that the conditional distribution $\rho (n|x)$ is skewed to the

1647:  right,

1648:  where the bins are chosen equally spaced on a logarithmic scale as

1649:  $x\in [10^{k-0.5},10^{k+0.5})$ with $k=3,\ldots,6$.

1650:  The dashed line shows a power-law fit to scatter plot data,

1651:  suggesting that the typical number of fishers scales with country

1652:  landings as $x^{\alpha}$.

1653:  From the slope of the dashed line, the exponent $\alpha$ is evaluated at

1654:  $0.68 \pm 0.08$.

1655: %

1656:  ({\sf B})~Stretched exponential law in the conditional probability of

1657:  $n$ given $x$, $\rho (n|x)$.

1658:  The negative logs of complementary CDF,

1659:  $-\ln P(\geq n|x)$,

1660:  are plotted versus decimal log of the number of fishers,

1661:  $\log_{10} n$,

1662:  on a semi-logarithmic scale.

1663:  The data points fall onto straight lines.

1664:  The countries are partitioned into four groups

1665:  according to landings (in tonnes):

1666:  $10^{2.5} \leq x < 10^{3.5}\; (\lozenge)$,

1667:  $10^{3.5} \leq x < 10^{4.5}\; (\vartriangle)$,

1668:  $10^{4.5} \leq x < 10^{5.5}\; (\blacklozenge)$,

1669:  $10^{5.5} \leq x < 10^{6.5}\; (\blacktriangle)$.

1670:  The solid lines are stretched exponential fits to the data,

1671:  representing power-law curves,

1672:  $-\ln P(\geq n|x) = (n/{n_0})^{\alpha}$,

1673:  where

1674:  $\alpha = 0.86\; (\lozenge)$,

1675:  $0.72\; (\vartriangle)$,

1676:  $0.66\; (\blacklozenge)$ and

1677:  $0.60\; (\blacktriangle)$,

1678:  estimated from

1679:  Eqs.(\ref{eqn:moment-str-1}) and (\ref{eqn:moment-str-2}).

1680:  The estimated reference scales $n_0$ are depicted in Fig.{\sf A} with

1681:  symbol ($\circ$), which closely follow the dashed line.

1682:  The exponent characterizing stretched exponential distributions should

1683:  be the exponent $\alpha$ in the power-low dependence of the average

1684:  number $\overline{n}(x)$ of fishers in a country with annual domestic

1685:  landings $x$ as suggested in Fig.{\sf A}

1686:  (dashed line).

1687:  A power-law fit with $n_0 (x) = B x^{\alpha}$

1688:  to estimated $n_0$'s (solid line in Fig.{\sf A}) gives

1689:  $\alpha = 0.71 \pm 0.01$ and $B = 11 \pm 2$,

1690:  exhibiting a good agreement with values of the exponent estimated from

1691:  Eqs.(\ref{eqn:moment-str-1}) and (\ref{eqn:moment-str-2}).

1692: %

1693: \item[Fig.7]

1694:  Scaling in employment statistics for the period 1990--1996.

1695: %

1696:  ({\sf A})~To test the scaling hypothesis in

1697:  Eq.(\ref{eqn:fisher-scaling}),

1698:  the scaled probability distributions

1699:  $\rho_{\mbox{\scriptsize sc}}\;(=B x^{\alpha} \rho)$

1700:  are plotted

1701:  versus scaled number

1702:  $n_{\mbox{\scriptsize sc}}\;(=n/Bx^{\alpha})$ of fishers

1703:  (double-logarithmic scale).

1704:  The data for four different bins are scaled by using the values of

1705:  landings

1706:  $x=10^3\; (\lozenge)$,

1707:  $10^4\; (\vartriangle)$,

1708:  $10^5\; (\blacklozenge)$,

1709:  $10^6$ tonnes ($\blacktriangle$),

1710:  with $\alpha = 0.71$ and $B = 11$.

1711:  The conditional PDF $\rho (n|x)$ exhibits scaling, i.e.

1712:  the data collapse onto a single curve,

1713:  yielding a scaling function $f_n$.

1714:  The solid line depicts the stretched exponential distribution

1715:  [Eq.(\ref{eqn:stretched-exponential})].

1716: %

1717:  ({\sf B})~Shown is the data collapse of the four curves

1718:  in Fig.\ref{fig:6}{\sf B}.

1719:  The negative logs of complementary CDF,

1720:  $-\ln P(\geq n_{\mbox{\scriptsize sc}})$,

1721:  are plotted

1722:  versus scaled number $n_{\mbox{\scriptsize sc}}$ of fishers

1723:  (double-logarithmic scale).

1724:  The solid line depicts the power-law curve,

1725:  $-\ln P(\ge n_{\mbox{\scriptsize sc}}) =

1726:  n_{\mbox{\scriptsize sc}}^{\alpha}$.

1727: %

1728: \item[Fig.8]

1729:  Recovering a log-normal distribution from a stretched exponential.

1730:  Semi-logarithmic scale plot of

1731:  probability distribution of the number of fishers for all countries

1732:  (excluding China).

1733:  The abscissa is chosen as $\log_{10} n$.

1734:  The data points are the average over the period 1990--1996, and

1735:  the dashed line is a fit of the log-normal PDF to the

1736:  data with geometric mean

1737:  $\hat{n} = 12.9$ (in thousand fishers)

1738:  and geometric standard deviation

1739:  $s_n = 2.09$.

1740:  Numerically integrating the conditional PDF $\rho (n|x)$ using

1741:  stretched exponential function Eq.(\ref{eqn:stretched-exponential})

1742:  with $\alpha = 0.71$

1743:  over a PDF of the country landings using

1744:  Eq.(\ref{eqn:lognormal-landings}),

1745:  a log-normal distribution of the number of fishers is approximately

1746:  recovered (solid line).

1747:  The reference size of the stretched exponential distribution

1748:  $\rho (n|x)$ is $n_0 (x) = Bx^{\alpha}$ with $B = 11$.

1749:  The distribution of the landing data for years 1990--1996 has

1750:  the geometric mean

1751:  $\hat{x} = 17.1$ (in thousand tonnes) and

1752:  the geometric standard deviation

1753:  $s_x = 3.22$.

1754: %

1755: \item[Fig.9]

1756:  Statistics on the marketing of fishery products in landing

1757:  sites (203 ports) in Japan.

1758: %

1759:  ({\sf A})~Semi-logarithmic scale plot of probability distribution of

1760:  local capture production $\xi$.

1761:  The abscissa is chosen as logarithmically scaled landings

1762:  $\ln (\xi/\hat{\xi})/{s_{\xi}}$,

1763:  where geometric means $\hat{\xi}$

1764:  of landings (in thousand tonnes) and geometric standard deviations

1765:  $s_{\xi}$ are as follows:

1766:  $\hat{\xi} = 4.01$,

1767:  3.60 and

1768:  3.33, and

1769:  $s_{\xi} = 1.75$,

1770:  1.70 and

1771:  1.66,

1772:  from years

1773:  2000 $(\square)$,

1774:  2001 $(\blacksquare)$ and

1775:  2002 $(\lozenge)$, respectively.

1776:  Log-scaled landings fall onto

1777:  the standardized Gaussian distribution (dashed line).

1778:  The port landing data exhibit the log-normal probability distribution,

1779:  same as the distribution of country landings (Fig.\ref{fig:2}{\sf A}).

1780: %

1781:  ({\sf B})~Semi-logarithmic scale plot of probability density function

1782:  of annual growth rate $r_{\xi}$ of port landings.

1783:  The points are the average over the years 2000--2002.

1784:  We find the same tent shape distribution as

1785:  the distribution of country landings growth rate plotted on

1786:  a semi-logarithmic scale.

1787:  The solid line is an exponential fit,

1788:  $

1789:  p(r_{\xi}) =

1790:  (\sqrt{2}\sigma_{\xi})^{-1}

1791:  \exp \left(-\sqrt{2} |r_{\xi}-\overline{r}_{\xi}|/\sigma_{\xi}\right)

1792:  $,

1793:  with average growth rate $\overline{r}_{\xi} = -0.06$

1794:  and standard deviation $\sigma_{\xi} = 0.30\pm0.02$.

1795:  The dashed line is a Gaussian fit with standard deviation

1796:  $\sigma_{\mbox{\tiny Gauss}} = 0.49$.

1797: %

1798: \end{itemize}

1799: 

1800: \newpage

1801: \begin{flushleft}

1802: {\bf Figures}

1803: \end{flushleft}

1804: 

1805: \begin{itemize}

1806:  \setlength{\itemsep}{5pt}

1807:                  %

1808: %

1809:  \item[Fig.1]

1810:  World fisheries trend including ($\circ$)/excluding ($\bullet$) China.

1811:  \item[Fig.2]

1812:  Fisheries production and employment statistics.

1813:  \item[Fig.3]

1814:  Exploitation dynamics of fish stocks.

1815:  \item[Fig.4]

1816:  Scaling in stock exploitation dynamics.

1817:  \item[Fig.5]

1818:  Growth rates of country landings for longer time scales.

1819:  \item[Fig.6]

1820:  Statistical analysis of the number of fishers.

1821:  \item[Fig.7]

1822:  Scaling in employment statistics for the period 1990--1996.

1823:  \item[Fig.8]

1824:  Recovering a log-normal distribution from a stretched exponential.

1825:  \item[Fig.9]

1826:  Statistics on the marketing of fishery products in landing

1827:  sites in Japan.

1828: %

1829: \end{itemize}

1830: 

1831: \newpage

1832: %

1833: % Fig.1

1834: \begin{figure}[p]

1835:  \centering

1836:  \includegraphics[width=6.5cm]{./fig1A.eps}

1837:  \hspace{0.5cm}

1838:  \includegraphics[width=6.5cm]{./fig1B.eps}

1839:  \caption{}

1840:  \label{fig:1}

1841: \end{figure}

1842: %

1843: 

1844: % Fig.2

1845: \begin{figure}[p]

1846:  \centering

1847:  \includegraphics[width=6.5cm]{./fig2A.eps}

1848:  \hspace{0.5cm}

1849:  \includegraphics[width=6.5cm]{./fig2B.eps}

1850:  \caption{}

1851:  \label{fig:2}

1852: \end{figure}

1853: %

1854: 

1855: % Fig.3

1856: \begin{figure}[p]

1857:  \centering

1858: {

1859:  \includegraphics[width=6.5cm]{./fig3A.eps}

1860:  \hspace{0.5cm}

1861:  \includegraphics[width=6.5cm]{./fig3B.eps}

1862: }

1863:  \caption{}

1864:  \label{fig:3}

1865: \end{figure}

1866: %

1867: 

1868: % Fig.4

1869: \begin{figure}[p]

1870:  \centering

1871: {

1872:  \includegraphics[width=6.5cm]{./fig4A.eps}

1873:  \hspace{0.5cm}

1874:  \includegraphics[width=6.5cm]{./fig4B.eps}

1875: }

1876:  \caption{}

1877:  \label{fig:4}

1878: \end{figure}

1879: 

1880: % Fig.5

1881: \begin{figure}[p]

1882:  \centering

1883: {

1884:  \includegraphics[width=6.5cm]{./fig5A.eps}

1885:  \hspace{0.5cm}

1886:  \includegraphics[width=6.5cm]{./fig5B.eps}

1887: }

1888: \vspace{0.0cm}

1889: {

1890:  \includegraphics[width=6.5cm]{./fig5C.eps}

1891:  \hspace{0.5cm}

1892:  \includegraphics[width=6.5cm]{./fig5D.eps}

1893: }

1894:  \caption{}

1895:  \label{fig:5}

1896: \end{figure}

1897: %

1898: 

1899: % Fig.6

1900: \begin{figure}[p]

1901:  \centering

1902: {

1903:  \includegraphics[width=6.5cm]{./fig6A.eps}

1904:  \hspace{0.5cm}

1905:  \includegraphics[width=6.5cm]{./fig6B.eps}

1906: }

1907:  \caption{\small

1908: }

1909:  \label{fig:6}

1910: \end{figure}

1911: %

1912: 

1913: % Fig.7

1914: \begin{figure}[tb]

1915:  \centering

1916: {

1917:  \includegraphics[width=6.5cm]{./fig7A.eps}

1918:  \hspace{0.5cm}

1919:  \includegraphics[width=6.5cm]{./fig7B.eps}

1920: }

1921:  \caption{}

1922:  \label{fig:7}

1923: \end{figure}

1924: %

1925: 

1926: % Fig.8

1927: \begin{figure}[tb]

1928:  \centering

1929: {

1930:  \includegraphics[width=6.5cm]{./fig8.eps}

1931: }

1932:  \caption{}

1933:  \label{fig:8}

1934: \end{figure}

1935: %

1936: 

1937: % Fig.9

1938: \begin{figure}[p]

1939:  \centering

1940: {

1941:  \includegraphics[width=6.5cm]{./fig9A.eps}

1942:  \hspace{0.5cm}

1943:  \includegraphics[width=6.5cm]{./fig9B.eps}

1944: }

1945:  \caption{}

1946:  \label{fig:9}

1947: \end{figure}

1948: %

1949: \end{document}

1950: