1: \documentclass{elsart}

2: \usepackage{natbib}

3: \usepackage{graphicx}

4: \usepackage{amsmath, amssymb}

5: 

6: \hyphenation{Misund Coetzee Vicsek Clupea harengus Pollachius}

7: 

8: %%% TeX Start

9: \begin{document}

10: 

11: \begin{frontmatter}

12: 

13: % Title, authors and addresses

14: 

15: \title{Power-law scaling in dimension-to-biomass relationship of fish

16:  schools}

17: 

18: 

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

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

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

22: %\author{Hiro-Sato Niwa}

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

24: 

25: \address{Behavioral Ecology Section,

26: National Research Institute of Fisheries Engineering,

27: Hasaki, Ibaraki 314-0421, Japan}

28: 

29: \begin{abstract}

30: Motivated by the finding that there is some biological universality in

31:  the relationship between school geometry and school biomass of various

32:  pelagic fishes in various conditions,

33:  I here establish a scaling law for school dimensions:

34:  the school diameter increases as a power-law function of school biomass.

35: The power-law exponent is extracted through the data collapse, and is

36:  close to $3/5$.

37: This value of the exponent implies that the mean packing density

38:  decreases as the school biomass increases, and

39: the packing structure displays a mass-fractal dimension of $5/3$.

40: By exploiting an analogy between school geometry and polymer chain

41:  statistics,

42: I examine the behavioral algorithm governing the swollen conformation of

43:  large-sized schools of pelagics, and

44:  I explain the value of the exponent.

45: \end{abstract}

46: 

47: \begin{keyword}

48: power-law scaling\sep

49: data collapse\sep

50: pelagic fish\sep

51: school size\sep

52: geometry 

53: \end{keyword}

54: 

55: \end{frontmatter}

56: 

57: % main text

58: %

59: 

60: \section{Introduction}

61: %

62: \subsection{Brief history of studies on the packing geometry of fish schools}

63: 

64: Animal packing in social aggregations is of fundamental interest in

65: ecology, and their conformations have been extensively studied

66: \citep{Parr27,Symons71,Okubo-Chiang74,Graves76,Pitcher-Partridge79,Aoki80,Partridge80,Partridge82,Dill81,Aoki-Inagaki88,misund93a,Parrish-Hamner97}.

67: Pioneer tank observations of pelagic fishes (herring, sprat and mackerel)

68: were conducted by \citet{Parr27}, and the ``persistently uniform

69: density'' of a school was noted.

70: From experiments on schools of saithe {\it Pollachius virens}, herring

71: {\it Clupea harengus} and cod {\it Gadus morhua} cruising in a ten

72: meters circular gantry tank,

73: \citet{Pitcher-Partridge79} validated that all the fish in a

74: school occupy a volume proportional to $Nb^3$,

75: where $N$ is the number of fish and $b$ is the mean nearest-neighbor

76: distance (approximate to the mean fish-body length) in the school.

77: \citet{misund93a} reported from field observations that 

78: the number densities of herring schools are almost independent of the

79: dimensional size of school

80: but are an order of magnitude lower than the packing densities observed

81: when schooling in small tanks.

82: 

83: Tauti and colleagues

84: \citep{Tauti-Miyosi29,Tauti-Hudino29,Tauti-Yasuda29,Tauti-Yasuda30,Tauti-Yasuda33a,Tauti-Yasuda33b}

85: had experimentally shown that a fish school can be viewed as a group of

86: inorganic particles and treated with the methods of physics.

87: \citet{breder54} explicitly applied physical equations to such

88: fish schools.

89: Recently,

90: a number of theoretical and numerical models of schooling and flocking

91: behavior have been studied by biologists, mathematicians and physicists

92: \citep{okubo-levin01,camazine-etal01,vicsek01-book}.

93: As for the animal group geometry,

94: many models for social aggregations, however, predict that densities

95: increase as the group size (in number) increases

96: [overviews of such models are given in

97: Refs. \citep{Warburton-Lazarus91,Beecham-Farnsworth99}].

98: \citet{Mogilner-etal03} mathematically tackled this problem of constant

99: density and revealed the condition for a well-spaced group, i.e.

100: what class of mutual interactions results in a relatively constant

101: individual distance in the interior of the aggregate.

102: 

103: More recently, by means of underwater acoustics,

104: the school sizes (in number or biomass of fish) have been quantitatively

105: measured for different values of the dimensional size of schools in the

106: wild.

107: Precise data on conformations of large-sized schools of pelegic fishes

108: became available rapidly,

109: which were extremely helpful in elucidating a certain geometric law, i.e.

110: the relation between dimensional and biomass sizes of pelagic fish

111: schools,

112: bringing about some important changes in our viewpoints.

113: Misund and

114: colleagues \citep{Misund90,Misund-etal92,Misund93b,Misund-etal95,Misund-etal96,Coetzee00,Misund-Coetzee00,Misund-etal03}

115: found that the power-law scaling in dimension-to-biomass

116: relationship exists, and is robust across a broad range of pelagic species

117: as well as across diverse environments.

118: They demonstrated that 

119: if the biomass $N$ in a school is, say, doubled,

120: the cross-sectional area of a school is increased by a

121: factor $2^{2\nu}$, i.e.

122: \begin{equation}

123:  \mbox{cross-sectional area} \propto (\mbox{biomass})^{2\nu},

124:   \label{eqn:misund-school-geometry}

125: \end{equation}

126: and that the exponent $\nu$ looks universal,

127: reading $0.5$

128: (from the field data $\nu$ ranges from $0.415$ to $0.77$).

129: The geometric law they found implies that

130: the mean density of a school scales as $N^{1-3\nu}$ in three dimensions

131: of space and the conformation of social aggregations swells

132: (i.e. $\nu > 1/3$).

133: Such a relationship as Eq.(\ref{eqn:misund-school-geometry}) with

134: $\nu = 0.5$ has been utilized for the dimensions to biomass conversion

135: (e.g. transforming the school diameter $R$ to the number $N$ of fish)

136: in analyzing school-size distributions

137: \citep{Anderson81,Niwa96b,Niwa98,Niwa03,Niwa04a}.

138: 

139: It might come as a surprise that

140: packing densities decrease as the group size increases,

141: contrary to previous observations and predictions.

142: Laboratory observations for school geometry and internal structure have

143: been made exclusively in small tanks,

144: which generally show the constant density of fish that the school

145: volume is proportional to the number of individuals.

146: In all set-ups, the factors resulting in homogeneous, cohesive school

147: may be especially pronounced, and

148: it therefore seems that the quantified structure is skewed.

149: \citet{Pitcher-Parrish93} claimed that homogeneity in schools has been

150: over-emphasized.

151: {\it In-situ} observations of herring and sprat schools with a high

152: resolution sonar revealed that the packing structure within the schools

153: is rather heterogeneous \citep{Cushing77}.

154: This has been confirmed by measurements of free-swimming schools using

155: photography and high-resolution echo integration, which showed that the

156: packing density distribution in capelin {\it Mallotus villosus} and

157: clupeoid schools varies considerably \citep{Freon-etal92,misund93a}.

158: Regions of high density are usually found within the schools, and even

159: empty vacuoles have been recorded.

160: \citet{Misund-Floen93} observed by repeated echo integration that there

161: were large variations in internal packing density of herring schools

162: (i.e. high-density regions or empty lacunas within a school),

163: and that the packing density structure was quasi-stationary.

164: Besides artificial environments in small tanks,

165: the discrepancy in former observations could have been caused by too

166: small numbers of fish in the schools.

167: For instance, in \citet{Pitcher-Partridge79},

168: $N$ takes a few tens of fish.

169: Since such a geometric law above is always defined only in a certain

170: limit \citep{deGennes79},

171: the scaling in the dimension-to-biomass relationship is expected to hold

172: for large-sized schools of pelagics.

173: 

174: In this paper, the exponent $\nu$ is estimated according to

175: the established universal scaling law in the school-size distribution

176: of pelagic fishes \citep{Niwa03,Niwa04a}:

177: choose the suitable value of $\nu$ to achieve the best data-collapse on

178: the size distributions in terms of the school dimension.

179: Notice that the dimension-to-biomass relationship is a property of the

180: single school and the scaling exponent $\nu$ is determined by the

181: behavioral algorithm of fish schooling at individual level,

182: while the scaling in the school-size distribution emerges from the

183: inter-school interactions at population level

184: (i.e. a global property of the interacting school system).

185: 

186: 

187: \subsection{Scaling in school-size distributions}

188: 

189: Animal group size is a focal issue in ecology that,

190: in contrast to scaling,

191: has introduced a single preferred size

192: (i.e. optimal or compromise size)

193: for any organism living in groups

194: \citep{Pulliam-Caraco84,Higashi-Yamamura93,Niwa96b,Hoare-etal04}.

195: Figure~\ref{fig:1} shows an example of the histogram of school

196: dimensions of Japanese sardine {\it Sardinops melanostictus}.

197: A peak frequency and a right skew are typical of pelagic fishes

198: \citep{Anderson81,Niwa96b,Niwa98}.

199: Their linear dimension, e.g. the vertical thickness, of the school

200: falls into a certain range below a few tens of meters.

201: One fish may not be the right atomic unit in schooling, since field

202: observations suggest that no school exists under a certain minimal size.

203: This may cause binned data of school sizes to exhibit the fake peak

204: frequency.

205: \citet{Anderson81} and \citet{Niwa96b} have ignored

206: slowly decaying (or fat-tailed) distributions, including scaling laws,

207: in such data of school sizes short ranged with a fake peak frequency

208: (the possibility of power laws was already presented in their models but

209: not exploited).

210: A traditional, widely used Gauss statistics says that, for the data from

211: \citet{Hara90} shown in Fig.\ref{fig:1},

212: finding sardine schools ranging from 18 to 20 meters in vertical

213: thickness should only occur about once every $10^9$ detections of

214: schools

215: [for detail, consult \citet{Niwa04a}].

216: In other words, it is not the real world!

217: Aquatic observations actually say that finding such schools occurs about

218: once every 500 detections.

219: The probability that such schools are found is $10^6$ times as large!

220: %

221: % Fig.1

222: \begin{figure}[tb]

223:  \centering

224:  \includegraphics[width=6.5cm]{./fig1.eps}

225:  \caption{

226:  Vertical dimension (in meters) of Japanese sardine schools.

227:  Data from the acoustic survey by \citet{Hara90} off southeastern

228:  Hokkaido in the summer, 1982.

229:  The same data are available in \citet{Hara84}.

230:  Small schools of vertical dimensions in the first bin

231:  ($<\mbox{two meters}$) may be stray ones.

232:  }

233:  \label{fig:1}

234: \end{figure}

235: %

236: 

237: The possibility of scaling in such distributions was found by

238: \citet{Bonabeau-Dagorn95}.

239: Fat-tailed distributions have been found to quite generally characterize

240: the size heterogeneity of pelagic fish schools

241: \citep{Niwa98,Bonabeau-etal98,Bonabeau-etal99}.

242: Only lately, \citet{Niwa03,Niwa04a} showed that the school-size

243: distribution of pelagic fishes has the universal functional form of a

244: power-law decay and a crossover towards the exponential distribution.

245: 

246: 

247: \subsection{Scaling and data collapse in ecology}

248: 

249: Scaling laws have been found to characterize many patterns

250: in ecological systems \citep{Azovsky00,Chave-Levin03,Halley-etal04}.

251: There is an empirical rule about the relationship between the areas

252: ($L$) of island and the numbers ($S$) of species on them:

253: whether counting birds, butterflies, plants or copepods in ponds, a

254: tenfold smaller area contains roughly half the species.

255: This can be fitted by a power-law function

256: \begin{equation}

257:  S(L)\propto L^z

258:   \label{eqn:species-area-rel}

259: \end{equation}

260: with the exponent $z\approx 0.25$

261: \citep{Preston62,MacArthur-Wilson67,Harte-etal99,May99}.

262: The species-area relationship is widely used in rough assessments of

263: likely future rates of species extinction

264: (because observed rates of tropical deforestation are equated to

265: loss of habitat area,

266: we can assess the annual production of species committed to

267: extinction).

268: So, a more secure understanding of such relationships has great

269: practical importance \citep{May99}.

270: \citet{Banavar-etal99} applied finite-size scaling

271: \citep{Fisher-Barber72,Binder-Heermann88}

272: to test the hypothesis of scaling invariance in the species-area

273: relationship resting on a model proposed by \citet{Harte-etal99} for the

274: species abundance distribution.

275: Following the conventional notation in the physics literature, the

276: system sizes are hereafter denoted by $L$.

277: 

278: The finite-size scaling (FSS) hypothesis assumes that

279: the fraction of objects with size $n$ for a finite system of size $L$ is

280: written, with a scaling exponent $A$, as

281: \begin{equation}

282:  P(n;L)

283:   =

284:   L^{-A} F\left(n/L^A\right),

285:   \label{eqn:FSS-hypothesis}

286: \end{equation}

287: where $F$ is a universal scaling function.

288: $P(n;L)$ reads, for instance,

289: the species-abundance distribution, which defines

290: the probability that any given species on a census patch of area $L$

291: has $n$ individuals,

292: where

293: $P(n;L)S(L)$ is the expected number of species with $n$ individuals.

294: $L^A$ is a measure of the width of the probability distribution

295: $P(n;L)$,

296: e.g. the mean or the standard deviation of the distribution.

297: Equation~(\ref{eqn:FSS-hypothesis}) expresses the principle that 

298: the behavior of the finite system, i.e. $P(n;L)$, is controlled by the

299: ratio $n/L^A$.

300: We test the FSS hypothesis by data collapsing:

301: when

302: $y = L^A P(n;L)$

303: is plotted versus

304: $x = n/L^A$,

305: if the distributions across different $L$'s fall on a single curve

306: (e.g.

307: collapsing distinct abundance distributions measured in different

308: areas and with different total numbers of individuals onto one scaling

309: curve),

310: then we should be able to identify a scaling function $F(x)$.

311: In order to determine the exponent $A$,

312: we try a best-fitting procedure such that the family of curves

313: $P(n;L)$ collapses onto a single curve as well as possible.

314: We then obtain other scaling exponents (e.g. $z$ of the species-area

315: relationship) resting on the scaling relation between scaling

316: exponents \citep{Goldenfeld92}.

317: By noting that

318: the total number of individuals of all species is equal to

319: $\sum_n n P(n;L)S(L)$,

320: the species-area relationship leads to a scaling relation

321: \begin{equation}

322:  A+z=1,

323:   \label{eqn:scaling-ecology}

324: \end{equation}

325: assuming that

326: the total number of individuals of all species is proportional

327: to area $L$ \citep{Banavar-etal99}.

328: 

329: The universal scaling is fundamental to data collapse.

330: The power-law scaling emerges as ubiquitous properties in ecology.

331: In statistical analysis in ecology, data collapsing across different

332: environments or species has been, however, observed only lately in

333: dynamics of breeding bird populations

334: \citep{Keitt-Stanley98,Keitt-etal02},

335: food web structure \citep{Camacho-etal02},

336: microbial body-mass spectra in marine ecosystems \citep{Rinaldo-etal02},

337: school-size distributions of pelagic fishes \citep{Niwa03,Niwa04a},

338: and in the context of ecological economics,

339: capture fisheries productions of countries \citep{Niwa04b}.

340: 

341: 

342: \section{Scaling in dimension-to-biomass relationship}

343: %

344: \subsection{Estimation of the exponent $\nu$ through data collapse}

345: 

346: In order to understand the geometric properties of school

347: configurations, the school biomass $N$ must be measured for different

348: values of dimensional size $R$, and we must compare them.

349: Misund and colleagues extensively performed the simultaneous

350: observations of the values of $R$ and $N$ for pelagics by the following

351: two methods:

352: (i)~two-dimensional (cross-sectional) acoustic measurements (unit in

353: square meters) and subsequent purse seine capture (unit in tonnes) of

354: schools \citep{Misund90,Misund93b},

355: and

356: (ii)~acoustic measurements of the three-dimensional structures and

357: backscattered echo energy \citep{Misund-etal92,Misund-etal95,Misund-etal96,Coetzee00,Misund-Coetzee00,Misund-etal03}.

358: The surveys were conducted on stocks of

359: anchovy {\it Engraulis capensis},

360: herring {\it Clupea harengus},

361: horse mackerel {\it Trachurus trachurus capensis},

362: mackerel {\it Scomber scombrus},

363: pilchard {\it Sardinops ocellatus},

364: round herring {\it Etrumeus whiteheadi},

365: saithe {\it Pollachius virens},

366: sardine {\it Sardinops sagax},

367: and

368: sprat {\it Sprattus sprattus},

369: in different seasons and geographic regions

370: (the Barents Sea,

371: the North Sea,

372: the Norwegian Sea,

373: the northeastern Atlantic and off Namibia,

374: and

375: off the coast of South Africa).

376: They found that there is some biological universality in

377: the dimensions-to-biomass relationships,

378: Eq.(\ref{eqn:misund-school-geometry}).

379: Their finding implies that

380: the radius $R$ of the school scales as

381: \begin{equation}

382:  R = (\mbox{constant})\times N^{\nu}

383:   \label{eqn:scaling-school-geometry}

384: \end{equation}

385: in a statistical sense.

386: $N$, denoting the school biomass, is hereafter defined by the number of

387: fish in a school.

388: 

389: Now I propose another way to establish the scaling law for school

390: dimensions, Eq.(\ref{eqn:scaling-school-geometry}), by applying

391: universal scaling law in fish school-biomass distributions

392: \citep{Niwa03,Niwa04a}.

393: The biomass distributions $W(N)$ follow a power law with exponent

394: $\beta =1$

395: up to a cut-off size

396: $\langle N \rangle_P$,

397: \begin{equation}

398:  W(N)

399:   =

400:   N^{-\beta} P \left(N/\langle N\rangle_P\right),

401:   \label{eqn:biomass-distribution}

402: \end{equation}

403: where $P(x)$ is a crossover scaling function with a strong drop for

404: $x>1$, and the cut-off size

405: (crossover size from power-law to exponential decay)

406: is calculated from the biomass histogram data

407:  $\{(N_i,W_i)|i=1,2,\ldots\}$,

408: \begin{equation}

409:  \langle N\rangle_P

410:   =

411:    \frac{

412:    \sum_i N_i^2 W_i\Delta N

413:    }{

414:    \sum_i N_i W_i\Delta N

415:    },

416:    \label{eqn:cut-off-size}

417: \end{equation}

418: where histogram bins are chosen with width $\Delta N$. 

419: 

420: The dimension data of fish schools are binned with width $\Delta R$,

421: giving the set of frequencies

422: $\left\{\left.\left(R_i,W^{\mbox{\tiny (G)}}_i\right)\right| i=1,2,\ldots\right\}$.

423: From Eqs.(\ref{eqn:scaling-school-geometry}) and

424: (\ref{eqn:biomass-distribution}),

425: the distribution of geometric dimensions of fish schools is represented

426: as

427: \begin{equation}

428:  W^{\mbox{\tiny (G)}}(R)

429:   = 

430:   R^{-1}

431:   P^{\mbox{\tiny (G)}}\left(R/{\langle R\rangle_P}\right),

432:   \label{eqn:dimension-distribution}

433: \end{equation}

434: where

435: \begin{equation}

436:  \langle R \rangle_P

437:   =

438:   \langle N \rangle_P^{\nu},

439: \end{equation}

440: and

441: \begin{equation}

442:  P^{\mbox{\tiny (G)}}(x)

443:   =

444:   P(x^{1/\nu}).

445: \end{equation}

446: Therefore, the school-dimension distribution follows a power law with

447: the same exponent ``$-1$'' as the school-biomass distribution.

448: The power-law distribution $W^{\mbox{\tiny (G)}}(R)$ is truncated at a

449: cut-off size $\langle R \rangle_P$,

450: which is also calculated from histogram data of school geometric

451: dimensions,

452: \begin{equation}

453:  \langle R\rangle_P

454:   =

455:   \left[

456:    \frac{

457:    \sum_i R_i^{2/\nu} W^{(\mbox{\tiny G})}_i \Delta R

458:    }{

459:    \sum_i R_i^{1/\nu} W^{(\mbox{\tiny G})}_i \Delta R

460:    }

461:  \right]^{\nu}.

462:   \label{eqn:depend-cutoff-nu}

463: \end{equation}

464: The following normalizations are adopted for the scaling function

465: $P^{\mbox{\tiny (G)}}(x)$ and

466: the histogram data of geometric dimensions of fish schools, because the

467: cut-off size $\langle N \rangle_P$ is proportional to the total number

468: of fish in the school system

469: [\citet{Niwa03,Niwa04a}; see also Eq.(\ref{eqn:cutoff-population})]:

470: \begin{equation}

471: \int_0^{\infty} x^{1/\nu-1} P^{\mbox{\tiny (G)}}(x)\mbox{d}x

472:  = 1,

473:  \label{eqn:dimension-distribution-normalization}

474: \end{equation}

475: and

476: \begin{equation}

477:  \sum_i R_i^{1/\nu} W^{\mbox{\tiny (G)}}_i \Delta R

478:   =

479:   \langle R\rangle_P^{1/\nu},

480: \end{equation}

481: respectively.

482: 

483: Fat-tailed school-size distributions are necessarily truncated because

484: the population is finite.

485: Since the size $\langle R\rangle_P$ depends on the exponent $\nu$

486: following Eq.(\ref{eqn:depend-cutoff-nu}), so that we can determine the

487: value of $\nu$ by evaluating the location of the cut-off in the

488: power-law distribution $W^{\mbox{\tiny (G)}}(R)$.

489: The ordinary least squares regressions might, however, lead to a

490: ``wrong'' estimation of the exponent

491: \citep{Niwa98,Bonabeau-etal98,Bonabeau-etal99}.

492: I make use of the data collapse to extract the ``right'' exponent.

493: From Eqs.(\ref{eqn:dimension-distribution}) and

494: (\ref{eqn:dimension-distribution-normalization}), when

495: $

496: y=W^{\mbox{\tiny (G)}}\langle R\rangle_P

497: $

498: is plotted against

499: $x=R/\langle R\rangle_P$

500: with correct parameter $\nu$,

501: all the empirical data should collapse onto each other.

502: The power-law exponent of school-dimension distributions, $\nu$, is

503: then evaluated through data collapse.

504: Let us search for the value  of $\nu$ that places all the points most

505: accurately on a single curve.

506: We use a set of histogram data of vertical dimension of Japanese sardine

507: {\it Sardinops melanostictus} schools, from 22 acoustic surveys by

508: \citet{Hara90} off southeastern Hokkaido for seven years, 1981--1987.

509: \citet{Hara86} reported that

510: Japanese sardine migrate as a huge-sized school in number from a few

511: hundreds of thousands to a few million of fish.

512: To obtain the best data collapse,

513: the $x$-axis is divided into bins (Fig.\ref{fig:2}a), and

514: for each bin two-dimensional variance

515: \begin{equation}

516:  \epsilon

517:   =

518:   \left(\sigma_x/\mu_x\right)^2

519:   +

520:   \left(\sigma_y/\mu_y\right)^2

521:   \label{eqn:2D-var}

522: \end{equation}

523: is calculated, where $\sigma_x$ and $\sigma_y$ denote the standard

524: deviation of the mean $\mu_x$ and $\mu_y$, respectively.

525: The parameter $\nu$ is then estimated at value that minimize the mean of

526: two-dimensional variance for the bins (Fig.\ref{fig:2}b).

527: The mean of two-dimensional variance,

528: $\overline{\epsilon}$, is a measure to determine the goodness of

529: collapse \citep{Lillo-etal02,Lillo-etal03,Niwa04a}.

530: A good data collapse can be obtained by using the value

531: $\nu \approx 0.6$.

532: The resulting plot of empirical school data is shown in

533: Fig.\ref{fig:2}a.

534: Experimentally fitting the parameter $\nu$ to achieve a good data

535: collapse, ``$3/5$'', is reminiscent of the Flory value of the

536: exponent in a power-law dependence of the coil radius of a polymer chain

537: (in three-dimensional solutions) on the degree of

538: polymerization \citep{Flory53,deGennes79}.

539: %

540: % Fig.2

541: \begin{figure}[tb]

542:  \centering

543:  \includegraphics[width=6.5cm]{./fig2a.eps}

544:  \hspace{0.5cm}

545:  \includegraphics[width=6.5cm]{./fig2b.eps}

546:  \caption{

547:  Data collapse to extract the exponent $\nu$.

548:  (a)~Scaled distribution of geometric dimensions of sardine schools.

549:  $y= W^{\mbox{\tiny (G)}}\langle R\rangle_P$

550:  is plotted versus

551:  $x= R/\langle R\rangle_P$

552:  with $\nu = 0.6$ on double-logarithmic scale.

553:  The bins are chosen equally spaced on a logarithmic scale as

554:  $x \in \left[10^{-1+k/5}, 10^{-1+(k+1)/5}\right)$ with

555:  $k=1,2,\ldots,6$.

556:  The rectangle in gray reads the interval $\mu_y\pm\sigma_y$,

557:  i.e. the error $\sigma_y$ on the mean $\mu_y$ (indicated by the slit)

558:  for each bin.

559:  The solid line is a prediction of the mean-field theory \citep{Niwa03}.

560:  (b)~The mean of two-dimensional variance,

561:  $\overline{\epsilon}$,

562:  versus the power-law exponent

563:  $\nu$.

564:  Although $\overline{\epsilon}$ shows noisy fluctuations, it takes a

565:  minimum around $\nu= 0.6$.

566:  Data from \citet{Hara90} are analyzed.

567: }

568:  \label{fig:2}

569: \end{figure}

570: %

571: 

572: Here

573: we see that the power-law regime of the distribution is too short,

574: which misled \citet{Anderson81} and \citet{Niwa96b} into overlooking the

575: power-law distributions of school sizes.

576: Notice that many power laws that appeared in the ecology literature

577: span less than two orders of magnitude of scale \citep{Halley-etal04}.

578: The power law range of too few scales is not unique to ecology;

579: the largest numbers of power laws reported in the physical science are

580: for small ranges \citep{Hamburger-etal96}.

581: 

582: 

583: \subsection{Retest of the FSS in school-biomass distributions}

584: 

585: The acoustic-survey data are converted into a school-biomass histogram

586: as follows

587: \begin{equation}

588:  W(N) \Delta N

589:   \propto

590:   W^{\mbox{\tiny (G)}}(R) R^{1/\nu -1} \Delta R.

591:   \label{eqn:biomass-dimension-convert}

592: \end{equation}

593: We now crosscheck the value of $\nu$ through finite-size scaling (FSS)

594: analysis of school-biomass distribution \citep{Niwa04a}.

595: Since the finite population size causes the truncation of power-law

596: distribution

597: $W(N) \propto N^{-\beta}$,

598: there is a well-defined quantity

599: \begin{equation}

600:  L

601:   =

602:   \frac{

603:   \sum_i N_i^{1+\beta} W_i\Delta N

604:   }{

605:   \sum_i N_i^{\beta} W_i\Delta N

606:   },

607:   \label{eqn:sys-size-school}

608: \end{equation}

609: which depends on the system population size.

610: In order to characterize the finite size effects,

611: FSS hypothesis is used:

612: the distribution function depends on $N$ only through the ratio $N/L^A$,

613: \begin{equation}

614:  W(N; L) \mbox{d}N

615:   =

616:   L^{-B}

617:   F \left(N/L^A\right) \mbox{d}\left(N/L^A\right),

618: \end{equation}

619: where $F(x)$ is a universal function independent of fish population size.

620: The prefactor $L^{-B}$ is required to ensure the normalization

621: \begin{equation}

622:  \sum_i N_i^{\beta} W_i\Delta N = 1,

623: \end{equation}

624: where $P(N)\;[\equiv N^{\beta} W(N)]$ now represents the probability

625: distribution of school-biomass sizes.

626: From the FSS hypothesis, it is expected that when

627: $W L^{A+B}$

628: is plotted against

629: $N/L^A$

630: with correct parameters $A$ and $B$

631: all the data collapse onto a single curve.

632: The power-law exponent of biomass distributions, $\beta$, is

633: then evaluated through FSS analysis.

634: The value of $B/A$ is the estimate of the power-law exponent

635: \begin{equation}

636:  \beta = \frac BA.

637: \end{equation}

638: Let us analyze a set of 22 acoustic-survey data of sardine schools (same

639: as Fig.\ref{fig:2}) converted into biomass histograms by using

640: Eq.(\ref{eqn:biomass-dimension-convert}) with $\nu =3/5$.

641: In a simultaneous best-fitting procedure (Fig.\ref{fig:3}),

642: a good data collapse can be obtained by using the values

643: $A \approx 1$ and $B \approx 1$.

644: The power-law exponent derived from the FSS collapse is $\beta \approx 1$. 

645: The resulting plot is shown in Fig.\ref{fig:3}a.

646: The school-biomass distribution follows a power-law decay with exponent

647: $-1$, and is truncated at the cut-off size of Eq.(\ref{eqn:cut-off-size}).

648: The FSS collapse confirms the scaling laws for school sizes,

649: Eq.(\ref{eqn:scaling-school-geometry}) with $\nu =3/5$ and

650: Eq.(\ref{eqn:biomass-distribution}) with $\beta =1$.

651: %

652: % Fig.3

653: \begin{figure}[tb]

654:  \centering

655:  \includegraphics[width=6.5cm]{./fig3a.eps}

656: \hspace{0.5cm}

657:  \includegraphics[width=6.5cm]{./fig3b.eps}

658:  \caption{

659:  FSS analysis of school-size data.

660:  (a)~FSS plot of the biomass distribution on double-logarithmic

661:  scale.

662:  Dimension data of sardine schools (same as Fig.\ref{fig:2}) are converted

663:  by Eq.(\ref{eqn:biomass-dimension-convert}) with $\nu =3/5$.

664:  Here $y= W L^{A + B}$

665:  is plotted versus

666:  $x= N/L^A$

667:  with $A =B =1$.

668:  Two-dimensional variances [same as Eq.(\ref{eqn:2D-var})] are

669:  calculated for bins chosen equally spaced on a logarithmic scale as

670:  $x \in \left[10^{-1+(k-0.5)/3}, 10^{-1+(k+0.5)/3}\right)$

671:  with $k=0,1,\ldots,5$.

672:  The rectangle in gray is same as Fig.\ref{fig:2}.

673:  The solid line is a prediction of the mean-field theory \citep{Niwa03}.

674:  (b)~The region of the $AB$-plane in which

675:  the minimum of the mean of two-dimensional variance exists.

676:  A measure of data collapse for scaling, i.e. the mean of

677:  two-dimensional variance, $\overline{\epsilon}$, takes a minimum

678:  $\overline{\epsilon}_{\mbox{\scriptsize min}}$ for the right choice of

679:  $(A, B)$.

680:  The minimum is found with the precision i.e. width of the minimum,

681:  $\Delta\epsilon = 10^{-3}$ in black region

682:  ($

683:  \Delta\epsilon/\overline{\epsilon}_{\mbox{\scriptsize min}}

684:  \approx 3.12\times 10^{-3}

685:  $).

686:  The values of the parameters lie in the intervals

687:  $A = 0.999\pm 0.008$ and $B = 1.003\pm 0.024$,

688:  and therefore

689:  $\beta = 1.004\pm 0.032$.

690: }

691:  \label{fig:3}

692: \end{figure}

693: %

694: 

695: Finite-size scaling techniques have been applied to ecology and shown to

696: predict scaling relations between scaling exponents

697: in relative abundance of species [Eq.(\ref{eqn:scaling-ecology})]

698: \citep{Banavar-etal99,Aji-Goldenfeld01},

699: in dynamics of breeding bird populations \citep{Keitt-etal02},

700: in biomass-size distributions of seston \citep{Rinaldo-etal02},

701: and in exploitation of fish stocks \citep{Niwa04b}.

702: In school-size statistics, we expect to have a scaling relation.

703: We now choose the normalization

704: \begin{equation}

705:  \sum_i N_i W_i\Delta N = \Phi,

706:   \label{eqn:norm-phi}

707: \end{equation}

708: where $\Phi$ denotes the total fish population in the school system

709: ($\sum_i W_i\Delta N$ gives the total number of schools).

710: Since Eq.(\ref{eqn:sys-size-school}) implies 

711: $L^A \propto \Phi^{\gamma}$ with a scaling exponent $\gamma$,

712: the FSS relation for the school-biomass distribution is written as

713: \begin{equation}

714:  W(N)

715:   =

716:   N^{-\beta} P(N/\Phi^{\gamma}).

717: \end{equation}

718: Accordingly, the normalization of Eq.(\ref{eqn:norm-phi}) yields the

719: scaling relation

720: \begin{equation}

721:  2 -\beta =1/\gamma.

722: \end{equation}

723: The best-fitting value in data collapse, $\beta\approx 1$, gives

724: $\gamma\approx 1$.

725: As a consequence,

726: the location of the cut-off in the power-law distribution of school

727: sizes simply reads

728: \begin{equation}

729:  \langle N \rangle_P

730:   \propto

731:   \Phi,

732:   \label{eqn:cutoff-population}

733: \end{equation}

734: which is verified by means of extensive numerical simulations

735: \citep{Niwa03,Niwa04a}.

736: 

737: 

738: \section{Behavioral algorithm of fish schooling}

739: %

740: \subsection{Gaussian model}

741: 

742: Let us now investigate cohesive motion of schools in a three-dimensional

743: space from the viewpoint of the behavioral algorithms which govern their

744: formation and dynamics:

745: attraction between neighbors maintains cohesion of the school;

746: a tendency to align with neighbors produces collective motion of the

747: school.

748: The minimal model of cohesion is a linkage of neighbors consisting of

749: harmonic spring, because the inter-fish distance follows a Gaussian

750: distribution \citep{Dill81,Partridge82,Niwa96a}.

751: Using the relative coordinates to the center of school,

752: the equation of motion of fish (as noisy self-propelled particles) in a

753: large school of size $N$ is written by a one-body approximation as the

754: following Langevin equation \citep{Niwa94,Niwa96a}:

755: \begin{equation}

756:  \frac{\mbox{d}^2\boldsymbol{r}}{\mbox{d}t^2}

757:   =

758:   \boldsymbol{f} (\boldsymbol{r})

759:   -J\frac{\mbox{d}\boldsymbol{r}}{\mbox{d}t}

760:   +\boldsymbol{\eta}(t),

761:   \label{eqn:langevin}

762: \end{equation}

763: providing that individuals are sufficiently sensitive to behavior of

764: their neighbors,

765: where $\boldsymbol{f}(\boldsymbol{r})$ is the cohesive force;

766: $J$ denotes the strength of alignment;

767: and

768: $\boldsymbol{\eta}$ is a random perturbation of the velocity with

769: strength $\epsilon_v$ and a $\delta$-correlated time dependence.

770: 

771: 

772: \subsection{Self-consistent calculation of of cohesive force}

773: 

774: The attractive interaction force acting on one body due to a system of

775: bodies is the neighbor-joining harmonic spring.  One essential

776: approximation is to replace the many-fish problem by the problem of

777: solving the motion of one fish in a certain self-consistent field.  The

778: cohesive force is then written in the following form by the one-body

779: approximation:

780: \begin{equation}

781:  \boldsymbol{f} (\boldsymbol{r})

782:   =

783:   -\frac{\tilde{k}}{N^{\alpha}} \boldsymbol{r},

784:   \label{eqn:cohesion}

785: \end{equation}

786: where $\tilde{k} N^{-\alpha}$ is the effective spring constant for the

787: springs strung out from the center of the school to a fish:

788: the number of consecutive springs joining the fish to the center

789: of the school via other companions is proportional to $N^{\alpha}$.

790: The total ``elastic energy'' of inter-fish bonds in the school depends

791: linearly on number $N$ of individuals in the school.

792: The overall elastic energy after integration over a sphere of radius $R$

793: (denoting the average radius of the school) results in

794: $

795:  E_{\mbox{\scriptsize el}}

796:   \propto

797:   {R^4}{N^{-\alpha}}

798: $,

799: which is derived from the ideal harmonic spring, Eq.(\ref{eqn:cohesion}).

800: Therefore, the exponent $\alpha$ will be determined self-consistently,

801: \begin{equation}

802:  \frac{R^4}{N^{\alpha}}

803:   \propto

804:   N.

805:   \label{eqn:self-consistent-spring}

806: \end{equation}

807: The solution of the Langevin equation~(\ref{eqn:langevin}) with

808: Eq.(\ref{eqn:cohesion}) takes the following asymptotic forms

809: \citep{hori77}:

810: \begin{equation}

811:  \sigma_v^2\,

812:   \left[\,

813:    \equiv

814:    \left\langle\left(

815: 		\mbox{d}\boldsymbol{r}/\mbox{d}t\right

816: 		)^2\right\rangle\right]

817:    \approx

818:    {3\epsilon_v}/J,

819:   \label{eqn:fluctuation-dissipation}

820: \end{equation}

821: \begin{equation}

822:  \langle\boldsymbol{r}^2\rangle

823:   \approx

824:   \frac{\sigma_v^2}{\tilde{k}}N^{\alpha},

825:   \label{eqn:mean-square-radius}

826: \end{equation}

827: where the root-mean-square

828: $\sqrt{\langle \boldsymbol{r}^2 \rangle}$

829: gives the average radius of the school, $R$.

830: From Eq.(\ref{eqn:self-consistent-spring}) together with

831: Eq.(\ref{eqn:mean-square-radius}),

832: the self-consistent value of the exponent $\alpha$ is obtained:

833: \begin{equation}

834: \alpha = 1.

835: \end{equation}

836: As a consequence, the self-consistent cohesive force reads

837: \begin{equation}

838:  \boldsymbol{f} (\boldsymbol{r})

839:   =

840:   -\frac{\sigma_v^2}{b^2 N}\boldsymbol{r},

841: \end{equation}

842: where $b$ denotes the effective inter-fish distance

843: (a constant independent of $N$):

844: $b^2\equiv \sigma_v^2/\tilde{k}$.

845: The average radius of the school is then given by

846: $bN^{1/2}$.

847: Note that

848: Eq.(\ref{eqn:fluctuation-dissipation}) is an example of a more general

849: principle called the fluctuation-dissipation theorem \citep{kubo66}.

850: 

851: Notice that the above model has been developed in the absence of the

852: excluded volume interactions.

853: Social cohesion by harmonic spring interactions between neighboring

854: fish reminds us of the classical picture of a polymer chain based on 

855: the bead-spring model \citep{Rouse53}.

856: The exponent $\nu$ close to the value $3/5$ may then be understood by

857: taking the excluded volume effect \citep{Doi-Edwards86}.

858: Such a school without excluded volume effect may be called the

859: ``Gaussian'' school in line with polymer physics.

860: 

861: 

862: \subsection{The excluded volume effect}

863: 

864: From Eq.(\ref{eqn:langevin}),

865: the probability of the school radius being between $r$ and $r+\mbox{d}r$ is

866: given by the following \citep{hori77}:

867: \begin{equation}

868:  \Psi_0 (r)

869:   =

870:   4\pi r^2

871:   \left(\frac 3{2\pi b^2N}\right)

872:   \exp\left(

873:        -\frac{3r^2}{2b^2N}

874:      \right)

875:   \label{eqn:prob-gaussian-radius}

876: \end{equation}

877: in a stationary state in three dimensions

878: (i.e. the position vector $\boldsymbol{r}$ follows a Gaussian

879: distribution).

880: The Gaussian school model considered above permits fish to occupy the

881: same region in space.

882: Of course this is a physical impossibility since each fish possesses its

883: own finite volume.

884: Therefore, in the ``excluded volume'' school, there are a number of

885: Gaussian school configurations which are disallowed due to the steric

886: effect.

887: Let $p(r)$ be the probability that a Gaussian school configuration, as

888: counted in Eq.(\ref{eqn:prob-gaussian-radius}), is also allowable under

889: the excluded volume condition.

890: We now calculate the probability that no overlaps occur when we place

891: $N$ fish within a region of volume ($\sim r^3$), which will lead to an

892: estimation for $p(r)$.

893: The approach is due to \citet{Doi96}.

894: Letting $w$ be the volume which is effectively excluded to one fish by the

895: presence of another ($w \lesssim b^3$), the probability that one

896: particular fish will not overlap with another is then given by

897: $(1-w/r^3)$.

898: Since there are $N(N-1)/2$ possible combinations of pairs, the

899: probability that no overlap occurs in all of these combinations is given

900: by

901: \begin{equation}

902:  p(r)

903:   =

904:   (1-w/r^3)^{N(N-1)/2}

905:   =

906:   \exp \left(

907: 	-\frac{wN^2}{2r^3}

908:        \right),

909:   \label{eqn:prob-no-overlap-1}

910: \end{equation}

911: where $r^3 \gg w$ and $N\gg 1$.

912: Therefore, the probability distribution of the school radius $r$ can

913: then be written as

914: \begin{equation}

915:  \Psi (r)

916:   =

917:   \Psi_0(r) p(r)

918:   \propto

919:   r^2

920:   \exp\left(

921:        -\frac{3r^2}{2b^2N}

922:        -\frac{wN^2}{2r^3}

923:       \right)

924:   \label{eqn:prob-excluded-radius}

925: \end{equation}

926: for the excluded volume school.

927: 

928: Both $\Psi_0(r)$ and $\Psi(r)$ have a maximum at certain values of $r$.

929: Let us estimate the average size of the school radius in each model by

930: calculating the positions of these maxima.

931: The maximum of $\Psi_0(r)$ occurs at

932: $R_0 = (2b^2N/3)^{1/2}$.

933: The maximum of $\Psi(r)$ occurs at $R$,

934: which satisfies the following equation obtained by differentiating the

935: logarithm of Eq.(\ref{eqn:prob-excluded-radius}):

936: \begin{equation}

937:  \left(\frac{R}{R_0}\right)^5

938:   -\left(\frac{R}{R_0}\right)^3

939:   =

940:   \frac{9\sqrt{6}}{16}\frac{w \sqrt{N}}{b^3}.

941:   \label{eqn:diff-log}

942: \end{equation}

943: If $N \gg 1$,

944: the second term on the left-hand side of

945: Eq.(\ref{eqn:diff-log}) can be neglected and hence

946: \begin{equation}

947:  R

948:   \simeq

949:   R_0

950:   \left(\frac{w\sqrt{N}}{b^3}\right)^{1/5}

951:   \propto

952:   N^{3/5}.

953:   \label{eqn:rouse-excluded-volume-effect}

954: \end{equation}

955: The exponent $3/5$ is the very value extracted through the data

956: collapse.

957: 

958: The scaling in the relationship between geometric dimensions and biomass

959: of pelagic fish schools is analogous to that developed in polymer

960: physics \citep{deGennes79}.

961: The result suggests that the dynamics are unexpectedly ``simple'' and

962: depend primarily on common cohesive motion in these animate and

963: inanimate systems.

964: Though the above is a very rough theory of school conformation with the

965: excluded volume effect,

966: the overall statistical properties do not depend on the details of the

967: model, which is a consequence of universality \citep{Stanley95}.

968: 

969: 

970: \subsection{Exploiting the analogy to a polymer chain model}

971: 

972: As indicated in the Introduction, large internal variations in packing

973: density occur within a school.

974: \citet{Freon-Misund99} pointed out that a source of substantial

975: variation in internal school structure is the formation of subgroups.

976: Such subgroups have been observed in saithe schools \citep{Partridge81},

977: and in schools of minnow {\it Phoxinus phoxinus} \citep{Pitcher73} and

978: herring \citep{Pitcher-Partridge79}.

979: Relatively independent movements of such clusters of individuals can

980: open up empty spaces and cause large variation in school volume.

981: 

982: The polymer-chain analogue of the subgroup in school conformation is the

983: ``blob'' \citep{deGennes79}:

984: the polymer chain behaves as a series of blobs.

985: Based on the blob concept, for small value of the excluded volume $w$, a

986: subgroup within a school, with a number of

987: $\lambda_{\mbox{\tiny B}}$ of fish, must be nearly Gaussian.

988: We see this from Eq.(\ref{eqn:prob-excluded-radius}) when we find no

989: effect of $w$ if $w\lambda_{\mbox{\tiny B}}^{1/2}/b^3 <1$.

990: There is a certain value of

991: $\lambda_{\mbox{\tiny B}}\;[\sim (b^3/w)^2]$,

992: beyond which excluded volume effects become important.

993: A single school will appear Gaussian at scales

994: $r < r_{\mbox{\tiny B}}$ where

995: \begin{equation}

996:  r_{\mbox{\tiny B}}

997:   \simeq

998:   b \lambda_{\mbox{\tiny B}}^{1/2}

999:   \simeq

1000:   b^4/w,

1001: \end{equation}

1002: while at scales $r > r_{\mbox{\tiny B}}$ it will show excluded volume

1003: effects.

1004: According to the ``blob'' approach, an $N$-sized school can be described

1005: as a cluster of $N/\lambda_{\mbox{\tiny B}}$ subgroups.

1006: Inside the subgroup

1007: the core repulsion (by excluded volume effects) is a weak perturbation

1008: leading to a Gaussian state.

1009: The school radius is then written as

1010: $R\simeq r_{\mbox{\tiny B}} (\lambda_{\mbox{\tiny B}}^{-1}N)^{\nu}$

1011: in the native state.

1012: 

1013: Notice the plasticity of school geometry and internal structure.

1014: The compact, dense packing, as observed in artificial environments like

1015: small tanks, may be caused by the very strong stress.

1016: When the school is confronted by danger such as predators, the

1017: inter-fish distance decreases rapidly and all the vacuoles within the

1018: school collapse quickly \citep{Freon-etal92}.

1019: This change may be closely related to the coil-globule transition in a

1020: polymer solution \citep{Ptitsyn-etal68,deGennes75}.

1021: 

1022: In real schools the nature of the short range interaction is quite

1023: complicated like van der Waals-type inter-molecular forces.

1024: Accordingly, all fish within a school interact via the two-body

1025: potential deformed from a parabolic potential of harmonic spring

1026: [grouping forces have been expressed as the gradient of a potential

1027: function in previous model studies \citep{Mogilner-etal03}].

1028: The potential change, denoted by $\Delta U_{ij}$, is given as the

1029: correction term to cohesive interaction potential between the $i$-th and

1030: the $j$-th fish:

1031: $\Delta U_{ij}$ will include steric effects and also may involve weak

1032: attractions.

1033: The correction factor to the probability distribution $\Psi(r)$ of the

1034: school radius $r$, in place of Eq.(\ref{eqn:prob-no-overlap-1}), is then

1035: given by the configuration integral

1036: \begin{equation}

1037:  p(r)

1038:   \simeq

1039:   r^{-3N}

1040:   \int\!\!\cdots\!\!\int

1041:   \exp \left(

1042: 	-J\sum_{i<j}

1043: 	\Delta U_{ij}/\epsilon_v

1044:       \right)

1045:   \mbox{d}\boldsymbol{x}_1\cdots\mbox{d}\boldsymbol{x}_N,

1046:   \label{eqn:prob-no-overlap}

1047: \end{equation}

1048: where

1049: $\boldsymbol{x}_i$ denotes the position vector of the $i$-th fish, and

1050: $\exp (-J\Delta U_{ij}/\epsilon_v)$

1051: is integrated over the school configuration space.

1052: We can estimate $p(r)$ by means of the cluster expansion

1053: \citep{Mayer-Mayer40,Uhlenbeck-Ford62}.

1054: Introducing a function of separation between pairs,

1055: $\boldsymbol{r}_{ij} =\boldsymbol{x}_j-\boldsymbol{x}_i$,

1056: \begin{equation}

1057:  \zeta_{ij}

1058:   =

1059:   \exp \left(-J\Delta U_{ij}/\epsilon_v \right) -1,

1060: \end{equation}

1061: we integrate Eq.(\ref{eqn:prob-no-overlap}) in the Mayer cluster

1062: expansion over diagrams up to the second irreducible cluster,

1063: yielding

1064: \begin{equation}

1065:  p(r)

1066:   \simeq

1067:   \exp\left(

1068:        -\frac{wN^2}{2r^3}

1069:        -\frac{w_{\vartriangle}N^3}{6r^6}

1070:        \right)

1071: \end{equation}

1072: with the excluded volume parameter

1073: \begin{equation}

1074:  w

1075:   =

1076:   -\int \zeta_{12} \mbox{d}\boldsymbol{r}_{12},

1077:   \label{eqn:second-virial}

1078: \end{equation}

1079: and the other parameter defined by

1080: \begin{equation}

1081:  w_{\vartriangle}

1082:   =

1083:   -\int\!\!\int

1084:   \zeta_{12} \zeta_{13} \zeta_{23}\;

1085:   \mbox{d}\boldsymbol{r}_{12}\mbox{d}\boldsymbol{r}_{13}.

1086: \end{equation}

1087: The actual configuration is decided by maximizing 

1088: the distribution function $\Psi(r)=\Psi_0(r)p(r)$,

1089: and this maximization condition leads to the equation determining the

1090: school radius $R$,

1091: \begin{equation}

1092:  \left(\frac{R}{R_0}\right)^5

1093:   -\left(\frac{R}{R_0}\right)^3

1094:   -\frac{27w_{\vartriangle}}{16 b^6} \left(\frac{R}{R_0}\right)^{-3}

1095:   =

1096:   \frac{9\sqrt{6}}{16}\frac{w \sqrt{N}}{b^3}.

1097:   \label{eqn:diff-log-2}

1098: \end{equation}

1099: Notice that, when the first irreducible cluster integral in Mayer's

1100: expansion is only taken into account, the configuration integral,

1101: Eq.(\ref{eqn:prob-no-overlap}), is evaluated by

1102: Eq.(\ref{eqn:prob-no-overlap-1}).

1103: In case that $\Delta U_{ij}$ consists of a hard core repulsion and a

1104: short-range weak attraction

1105: (i.e. at the distance of near-collision ($r\sim w^{1/3}$) the

1106: attraction gradient between fish becomes steeper),

1107: Eq.(\ref{eqn:second-virial}) is estimated as

1108: $w=w_0 (1-\epsilon_{\Theta}/\epsilon_v)$,

1109: expressing the dependence of $w$ on the velocity-fluctuation

1110: $\epsilon_v$ of fish,

1111: where $w_0$ and $\epsilon_{\Theta}$ are constants

1112: \citep{Doi-Edwards86}.

1113: During predation threats or laboratory observations,

1114: the value of excluded volume parameter $w$ will change sign from

1115: positive to negative, provided that 

1116: the velocity fluctuation $\epsilon_v$ is suppressed below a certain

1117: value $\epsilon_{\Theta}$.

1118: Decreasing $\epsilon_v$ below $\epsilon_{\Theta}$,

1119: the dimensional size of the school becomes much smaller than that of a

1120: Gaussian school, as depicted in Fig.\ref{fig:4}.

1121: According to Eq.(\ref{eqn:diff-log-2})

1122: the so called ``expansion factor'' $R/R_0$ is determined not by $w$ but

1123: by $wN^{1/2}$,  and so if $N$ is large only a small change in

1124: $\epsilon_v$ will cause a big change in dimensional size.

1125: For example, for school of $10^6$ fish, a small variation in

1126: $\epsilon_v$ will induce a dramatic change in the school radius.

1127: Equation~(\ref{eqn:diff-log-2}) states that

1128: when $-wN^{1/2}/b^3 \gg 1$, $R/R_0$ is proportional to

1129: $(-wN^{1/2})^{-1/3}$ and hence

1130: the solution is as follows:

1131: \begin{equation}

1132:  R

1133:   \simeq

1134:   R_0

1135:   \left(\frac{-wb^3\sqrt{N}}{w_{\vartriangle}}\right)^{-1/3}

1136:   \propto

1137:   N^{1/3},

1138:   \label{eqn:globule-state}

1139: \end{equation}

1140: which shows the closest packing of fish in a school, i.e.

1141: the constant density independent of the number of individuals.

1142: This change may be called the swollen-dense packing transition of

1143: schools.

1144: %

1145: % Fig.4

1146: \begin{figure}[tb]

1147:  \centering

1148:  \includegraphics[width=6.5cm]{./fig4.eps}

1149:  \caption{

1150:  Swollen-dense packing transition in the school geometry

1151:  [the solution of Eq.(\ref{eqn:diff-log-2})].

1152:  Here $R'=R/R_0$ is plotted as a function of a combined variable

1153:  $w'=(9\sqrt{6}/16)\; wN^{1/2}/b^3$.

1154:  Shown is the transition region $|w|N^{1/2}/b^3\approx 1$.

1155:  At $\epsilon_v = \epsilon_{\Theta}$ the excluded volume parameter $w$

1156:  equals $0$.

1157:  The repulsive excluded volume effect then balances the attractive

1158:  forces between fish, and the school behaves as a Gaussian school.

1159:  The property of the swollen-dense packing transition is quite different

1160:  in the two regions separated by the critical value

1161:  $w_{\vartriangle}/b^6 = 0.0228$.

1162:  The transition between the the swollen conformation and the compact

1163:  packing is gradual when $w_{\vartriangle}/b^6 > 0.0228$ (dotted line),

1164:  while it becomes extremely sharp and the transition is discontinuous

1165:  when $w_{\vartriangle}/b^6 < 0.0228$ (gray line).

1166:  The solid line shows the marginal transition behavior, and 

1167:  the crossover takes place when $w' = -0.242$ (indicated by the arrow).

1168:  Notice that the solution of Eq.(\ref{eqn:diff-log-2}) is  multivalued

1169:  [depicted by a sigmoid curve (white cutout)], showing a  first-order

1170:  discrete phase change  \citep{Ptitsyn-etal68,deGennes75},

1171:  when $w_{\vartriangle}/b^6 < 0.0228$.

1172:  }

1173:  \label{fig:4}

1174: \end{figure}

1175: %

1176: 

1177: 

1178: \subsection{Configuration inside a swollen school}

1179: 

1180: The approach of treating fish schools as interacting particle systems

1181: naturally leads to the idea of applying successful methods of

1182: statistical physics to the description of moving together without a

1183: leader \citep{vicsek01}.

1184: In statistical physics, the presence of non-trivial scaling is usually

1185: taken to mean that the dynamics are largely governed by simple geometric

1186: properties of the system and do not depend strongly on detailed

1187: properties of the system components \citep{wilson83}.

1188: 

1189: As for the school geometry, the basic units are the effective inter-fish

1190: length $b$ and the number of fish, $N$.

1191: We assemble $\lambda$ neighboring fish in a school into $N/\lambda$

1192: groups, e.g. in the sense of the Voronoi-Dirichlet

1193: diagram \citep{dirichlet1850,voronoi08,voronoi09}.

1194: The length between centers of neighboring subgroups is defined by

1195: $\lambda^{\tilde\nu} b$ with an exponent $\tilde\nu$.

1196: After such a scale change as

1197: \begin{equation}

1198:  N\to \lambda^{-1}N,\quad

1199:   b\to \lambda^{\tilde\nu}b,

1200:   \label{eqn:scale-change}

1201: \end{equation}

1202: the macroscopic quantities which determine the overall properties of the

1203: school satisfy

1204: \begin{equation}

1205:  g(\lambda^{-1}N,\lambda^{\tilde\nu}b)

1206:   =

1207:   \lambda^{\chi} g(N,b),

1208: \end{equation}

1209: where $\chi$ is an exponent which depends on the quantity under

1210: consideration.

1211: For this to hold true for arbitrary $\lambda$, the function $g(N,b)$ must take

1212: the scaling form

1213: \begin{equation}

1214:  g(N,b)

1215:   =

1216:   N^{-\chi} g^{\mbox{\scriptsize (sc)}} (bN^{\tilde\nu}),

1217: \end{equation}

1218: where a new function $g^{\mbox{\scriptsize (sc)}}$ is introduced.

1219: This is the Widom-Kadanoff scaling law in critical

1220: phenomena \citep{Goldenfeld92}.

1221: For example, the average radius of the school, $R$,

1222: should be unaltered under the transformation of

1223: Eq.(\ref{eqn:scale-change}):

1224: \begin{equation}

1225:  R(N,b)

1226:   =

1227:   R(\lambda^{-1}N,\lambda^{\tilde\nu}b)

1228:   =

1229:   R^{\mbox{\scriptsize (sc)}}(bN^{\tilde\nu}).

1230: \end{equation}

1231: Because of Eq.(\ref{eqn:scaling-school-geometry}),

1232: $R^{\mbox{\scriptsize (sc)}} (x) = x$

1233: and we must have

1234: \begin{equation}

1235:  \tilde\nu = \nu,

1236: \end{equation}

1237: where $\nu = 3/5$ in three dimensions.

1238: 

1239: As an example of the application of the above scaling law,

1240: let us consider the pair correlation function inside a school,

1241: where fish movement take place in a three-dimensional space.

1242: A pair correlation function $\rho_2(r)$ is defined as follows.

1243: We pick one fish at random in the school, and we place it at origin.

1244: Then we ask, what is the number density of other fish at a (directional)

1245: distance $\boldsymbol{r}$ from the first, and we average the result over

1246: all choices of the first fish (and of directions).

1247: The function $\rho_2(r)$ has an integral which is just the total number

1248: of fish per school,

1249: $\displaystyle{\int \rho_2(r) \mbox{d}\boldsymbol{r} = N}$.

1250: Hence, from dimensional analysis, we can write

1251: \begin{equation}

1252:  \rho_2(r)

1253:   =

1254:   R^{-3} g(N,r/b).

1255: \end{equation}

1256: Under the transformation that neighboring $\lambda$ fish are grouped to

1257: form one subunit in a school,

1258: $\rho_2(r)$ will be reduced by $1/\lambda$, since the pair correlation

1259: function is proportional to the number density of fish.

1260: Therefore

1261: \begin{equation}

1262:  g(\lambda^{-1}N,r/\lambda^{\nu}b)

1263:   =

1264:   \lambda^{-1} g(N,r/b).

1265: \end{equation}

1266: Then, the function $\rho_2(r)$ obeys a simple scaling rule

1267: \begin{equation}

1268:  \rho_2(r)

1269:   =

1270:   N R^{-3}

1271:   \rho_2^{\mbox{\scriptsize (sc)}}(r/R),

1272:   \label{eqn:pair-correlation-scaling}

1273: \end{equation}

1274: where $\rho_2^{\mbox{\scriptsize (sc)}}$ is a dimensionless universal

1275: function, and $NR^{-3}$ is the dimensional

1276: factor.

1277: Focusing on the limit $r \ll R$, we can reach the form of

1278: $\rho_2(r)$ by a simple argument.

1279: In a sphere of radius $r$ we have a certain number of fish,

1280: $\tilde{N}$, related to $r$ by the excluded volume exponent:

1281: $\tilde{N}^{3/5} b \simeq r$.

1282: The function $\rho_2(r)$ scales like the density of fish in the

1283: sphere,

1284: \begin{equation}

1285:  \rho_2(r)

1286:   \simeq

1287:   \frac{\tilde{N}}{r^3}

1288:   \simeq

1289:   \frac 1{r^{4/3} b^{5/3}}

1290:   \quad

1291:   (\mbox{for}\,\, r<R),

1292:   \label{eqn:pair-correlation}

1293: \end{equation}

1294: which gives $\rho_2^{\mbox{\scriptsize (sc)}}(x) \simeq x^{-4/3}$.

1295: 

1296: The possible observable property,

1297: Eq.(\ref{eqn:pair-correlation-scaling}), tells us that if we were

1298: to measure $\rho_2(r)$ for fish schools of different size $N$,

1299: there would be superposition of the curves obtained by plotting

1300: $\rho_2(r)R^3 /N$ against $r/R$.

1301: Thus this kind of scaling relation may be verified experimentally on

1302: large schools of pelagic fish.

1303: 

1304: 

1305: \section{Discussion}

1306: 

1307: The power-law scaling generally exists in dimension-to-biomass

1308: relationship of pelagic fish schools in nature.

1309: Here I have estimated the power-law exponent $\nu$ of the geometric

1310: relation, based on the dependence of the power-law regime of the

1311: school-dimension distribution $W^{\mbox{\tiny (G)}}(R)$ on $\nu$,

1312: i.e. Eq.(\ref{eqn:depend-cutoff-nu}).

1313: We have tested whether the distributions $W^{\mbox{\tiny (G)}}(R)$ are

1314: self-similar (i.e. exhibit scaling), relying on both the power law for

1315: the dimension-to-biomass relationship of

1316: Eq.(\ref{eqn:scaling-school-geometry}) and the FSS relation in the

1317: power-law school-biomass distribution of

1318: Eq.(\ref{eqn:biomass-distribution}).

1319: Plotting the scaled histogram-data from the 22 {\it in-situ}

1320: observations, we have found that the 22 curves do indeed collapse onto

1321: each other (Fig.\ref{fig:2}a), suggesting that $W^{\mbox{\tiny (G)}}(R)$

1322: follows a universal functional form

1323: [Eq.(\ref{eqn:dimension-distribution})].

1324: We have extracted the power-law exponent of dimension-to-biomass scaling

1325: relationship via a minimization of a measure to quantify the nature of

1326: scaling collapse.

1327: Next I have explained the value of the exponent $\nu$, proposing the

1328: Gaussian school model for the fish with excluded volume $w$.

1329: By exploiting the analogy between fish-school and polymer

1330: conformations, we have examined the behavioral algorithm governing the

1331: swollen conformation of large-sized schools.

1332: We have seen that the exponent is modified strongly by the steric

1333: effect.

1334: What is universal in Eq.(\ref{eqn:scaling-school-geometry}) is the

1335: exponent $\nu\approx 0.6$:

1336: it is independent of species as well as environmental conditions, and

1337: the same for all schools.

1338: The constant that multiplies $N^{\nu}$ in

1339: Eq.(\ref{eqn:scaling-school-geometry}) is non-universal

1340: [$\sim (wb^2)^{1/5}$ with effective inter-fish distance $b$, predicted

1341: from Eq.(\ref{eqn:rouse-excluded-volume-effect})],

1342: and depends on the details of interactions between fish which may vary

1343: with species and environmental conditions \citep{Morgan88}.

1344: 

1345: To understand the scaling law for school geometry,

1346: Eq.(\ref{eqn:scaling-school-geometry}) with $\nu =3/5$,

1347: it is essential to see what the value of $\nu$ represents.

1348: Obviously, for a regular object embedded into a $d$-dimensional

1349: Euclidean space, Eq.(\ref{eqn:scaling-school-geometry}) would have the

1350: form $N(R)\sim R^d$ expressing the fact that the volume of a

1351: $d$-dimensional object grows with its linear size $R$ as $R^d$.

1352: Contrary to an integer dimensionality, it has been shown that the

1353: packing structure within the schools is characterized by a non-integer

1354: (i.e. fractal) dimensionality of $1/\nu \approx 1.7$:

1355: the number of fish in a school of radius $R$ scales as

1356: $N(R)\sim R^{1/\nu}$.

1357: During the last decades of the twentieth century

1358: it has widely been recognized by researchers working in diverse areas of

1359: science that many of the structures commonly observed possess a rather

1360: special kind of geometrical complexity;

1361: the name ``fractal'' was coined by Benoit Mandelbrot

1362: for these complex shapes \citep{Mandelbrot82}.

1363: Objects of biological origin are many times fractal-like

1364: \citep{vicsek01-book}.

1365: 

1366: FSS plots of data lead to possibly important and

1367: practical insights in ecology.

1368: In the context of the fisheries mission,

1369: the demonstrated geometric relation between dimensions and biomass of

1370: pelagic fish schools, when applied to mapping pelagic schooling fish,

1371: will largely improve the precision in the fish stock assessment.

1372: Because of the linear dependence of the fish population on

1373: $\langle N \rangle_P$,

1374: as given by Eq.(\ref{eqn:cutoff-population}),

1375: the fish stock abundance can be inferred from an index

1376: (i.e. cut-off size) that can be determined directly from observations

1377: [see Eq.(\ref{eqn:cut-off-size})].

1378: 

1379: % Acknowledgements.

1380: I am very grateful to Leah Edelstein-Keshet for helpful suggestions.

1381: 

1382: %REFERENCES

1383: \begin{thebibliography}{99}

1384: %

1385:  \bibitem[Aji and Goldenfeld(2001)]{Aji-Goldenfeld01}

1386: 	 Aji, V., Goldenfeld, N., 2001.

1387: 	 Fluctuations in finite critical and turbulent systems.

1388: 	 {Phys. Rev. Lett.} {86}, 1007--1010.,

1389: 	 doi:10.1103/PhysRevLett.86.1007.

1390: %

1391:  \bibitem[Anderson(1981)]{Anderson81}

1392: 	 Anderson, J. J., 1981.

1393: 	 A stochastic model for the size of fish schools.

1394: 	 {U.S. Fish. Bull.} {79}, 315--323.

1395: %

1396:  \bibitem[Aoki(1980)]{Aoki80}

1397: 	 Aoki, I., 1980.

1398: 	 An analysis of the schooling behavior of fish:

1399: 	 internal organization and communication process.

1400: 	 {Bull. Ocean Res. Inst. Univ. Tokyo} {12}, 1--65.

1401: %

1402:  \bibitem[Aoki and Inagaki(1988)]{Aoki-Inagaki88}

1403: 	 Aoki, I., Inagaki, T., 1988.

1404: 	 Photographic observations on the behavior of Japanese anchovy

1405: 	 {\it Engraulis japonica} at night in the sea.

1406: 	 {Mar. Ecol. Prog. Ser.} {43}, 213--221.

1407: %

1408:  \bibitem[Azovsky(2000)]{Azovsky00}

1409: 	 Azovsky, A. I., 2000.

1410: 	 Concept of scale in marine ecology: linking the words or the worlds?

1411: 	 {Web Ecol.} {1}, 28--34.

1412: %

1413:  \bibitem[Banavar {et al.}(1999)]{Banavar-etal99}

1414: 	 Banavar, J. R., Green, J. L., Harte, J., Maritan, A., 1999.

1415: 	 Finite size scaling in ecology.

1416: 	 {Phys. Rev. Lett.} {83}, 4212--4214.

1417: %

1418:  \bibitem[Beecham and Farnsworth(1999)]{Beecham-Farnsworth99}

1419: 	 Beecham, J. A., Farnsworth, K. D., 1999.

1420: 	 Animal group forces resulting from predator avoidance and

1421: 	 competition minimization.

1422: 	 {J. Theor. Biol.} {198}, 533--548.

1423: %

1424:  \bibitem[Binder and Heermann(1988)]{Binder-Heermann88}

1425: 	 Binder, K., Heermann, D. W., 1988.

1426: 	 {Monte Carlo Simulation in Statistical Physics}.

1427: 	 Springer-Verlag, Berlin.

1428: %

1429:  \bibitem[Bonabeau and Dagorn(1995)]{Bonabeau-Dagorn95}

1430: 	 Bonabeau, E., Dagorn, L., 1995.

1431: 	 Possible universality in the size distribution of fish schools.

1432: 	 {Phys. Rev. E} {51}, R5220--R5223.

1433: %

1434:  \bibitem[Bonabeau {et al.}(1998)]{Bonabeau-etal98}

1435: 	  Bonabeau, E., Dagorn, L., Fr\'eon, P., 1998.

1436: 	  Space dimension and scaling in fish school-size distributions.

1437: 	  {J. Phys. A: Math. Gen.} {31}, L731--L736.

1438: %

1439:  \bibitem[Bonabeau {et al.}(1999)]{Bonabeau-etal99}

1440: 	 Bonabeau, E., Dagorn, L., Fr\'eon, P., 1999.

1441: 	 Scaling in animal group-size distribution.

1442: 	 {Proc. Natl. Acad. Sci. USA} {96}, 4472--4477.

1443: %

1444:  \bibitem[Breder(1954)]{breder54}

1445: 	 Breder, Jr., C. M., 1954.

1446: 	 Equation descriptive of fish schools and other animal aggregations.

1447: 	 {Ecology} {35}, 361--370.

1448: %

1449:  \bibitem[Camacho {et al.}(2002)]{Camacho-etal02}

1450: 	 Camacho, J., Guimer\`a, R., Amaral, L. A. N., 2002.

1451: 	 Robust patterns in food web structure.

1452: 	 {Phys. Rev. Lett.} {88}, 228102.,

1453: 	 doi:10.1103/PhysRevLett.88.228102. 

1454: %

1455:  \bibitem[Camazine {et al.}(2001)]{camazine-etal01}

1456: 	 Camazine, S., Deneubourg, J.-L., Franks, N. R., Sneyd, J.,

1457: 	 Theraulaz, G., Bonabeau, E., 2001.

1458: 	 {Self-Organization in Biological Systems}.

1459: 	 Princeton University Press, Princeton.

1460: %

1461:  \bibitem[Chave and Levin(2003)]{Chave-Levin03}

1462: 	 Chave, J., Levin, S., 2003.

1463: 	 Scale and scaling in ecological and economic systems.

1464: 	 {Environ. Resource Econ.} {26}, 527--557.

1465: %

1466:  \bibitem[Coetzee(2000)]{Coetzee00}

1467: 	 Coetzee, J., 2000.

1468: 	 Use of a shoal analysis and patch estimation system (SHAPES) to

1469: 	 characterise sardine schools.

1470: 	 {Aquat. Living Resour.} {13}, 1--10.

1471: %

1472:  \bibitem[Cushing(1977)]{Cushing77}

1473: 	 Cushing, D. H., 1977.

1474: 	 Observations of fish schools with the ARL scanner.

1475: 	 {Rapp. P.-v R\'eun. Cons. int. Explor. Mer} {170}, 15--20.

1476: %

1477:  \bibitem[de Gennes(1975)]{deGennes75}

1478: 	 de Gennes, P.-G., 1975.

1479: 	 Collapse of a polymer chain in poor solvents.

1480: 	 {J. de Physique} {36}, L55--L57.

1481: %

1482:  \bibitem[de Gennes(1979)]{deGennes79}

1483: 	 de Gennes, P.-G., 1979.

1484: 	 {Scaling Concepts in Polymer Physics}.

1485: 	 Cornell University Press, Ithaca, New York.

1486: %

1487: \bibitem[Dill {et al.}(1981)]{Dill81}

1488: 	 Dill, L. M., Dunbrack, R. L., Major, P. F., 1981.

1489: 	 A new stereophotographic technique for analyzing the

1490: 	 three-dimensional structure of fish schools.

1491: 	 {Env. Biol. Fish.} {6}, 7--13.

1492: %

1493:  \bibitem[Dirichlet(1850)]{dirichlet1850}

1494: 	 Dirichlet, G. L., 1850.

1495: 	 {\"U}ber die {R}eduktion der positiven quadratischen {F}ormen

1496: 	 mit drei unbestimmten ganzen {Z}ahlen.

1497: 	 {J. Reine Angew. Math.} {40}, 209--227.

1498: %

1499:  \bibitem[Doi(1996)]{Doi96}

1500: 	 Doi, M., 1996.

1501: 	 {Introduction to Polymer Physics}.

1502: 	 Oxford University Press, New York.

1503: %

1504:  \bibitem[Doi and Edwards(1986)]{Doi-Edwards86}

1505: 	 Doi, M., Edwards, S. F., 1986.

1506: 	 {The Theory of Polymer Dynamics}.

1507: 	 Oxford University Press, New York.

1508: %

1509:  \bibitem[Fisher and Barber(1972)]{Fisher-Barber72}

1510: 	 Fisher, M. E., Barber, M. N., 1972.

1511: 	 Scaling theory for finite-size effects in the critical region.

1512: 	 {Phys. Rev. Lett.} {28}, 1516--1519.

1513: %

1514:  \bibitem[Flory(1953)]{Flory53}

1515: 	 Flory, P., 1953.

1516: 	 {Principles of Polymer Chemistry}.

1517: 	 Cornell University Press, Ithaca, New York.

1518: %

1519:  \bibitem[Fr{\'e}on {et al.}(1992)]{Freon-etal92}

1520: 	 Fr{\'e}on, P., Gerlotto, F., Soria, M., 1992.

1521: 	 Changes in school structure according to external stimuli:

1522: 	 description and influence on acoustic assessment.

1523: 	 {Fish. Res.} {15}, 45--66.

1524: %

1525:  \bibitem[Fr{\'e}on and Misund(1999)]{Freon-Misund99}

1526: 	 Fr{\'e}on, P., Misund, O. A., 1999.

1527: 	 {Dynamics of Pelagic Fish Distribution and Behaviour: Effects on

1528: 	 Fisheries and Stock Assessment}.

1529: 	 Fishing News Books, Blackwell, Osney Mead, Oxford.

1530: %

1531:  \bibitem[Goldenfeld(1992)]{Goldenfeld92}

1532: 	 Goldenfeld, N., 1992.

1533: 	 {Lectures on Phase Transitions and the Renormalization

1534: 	 Group}.

1535: 	 Vol.85 of {Frontiers in Physics}.

1536: 	 Addison-Wesley, Reading, Massachusetts.

1537: %

1538:  \bibitem[Graves(1976)]{Graves76}

1539: 	 Graves, J., 1976.

1540: 	 Photographic method for measuring spacing and density within

1541: 	 pelagic fish schools at sea.

1542: 	 {U.S. Fish. Bull.} {75}, 230--234.

1543: %

1544:  \bibitem[Hamburger {et al.}(1996)]{Hamburger-etal96}

1545: 	 Hamburger, D. A., Biham, O., Avnir, D., 1996.

1546: 	 Apparent fractality emerging from models of random

1547: 	 distributions.

1548: 	 {Phys. Rev. E} {53}, 3342--3358.

1549: %

1550:  \bibitem[Halley {et al.}(2004)]{Halley-etal04}

1551: 	 Halley, J. M., Hartley, S., Kallimanis, A. S., Kunin, W. E.,

1552: 	 Lennon, J. J., Sgardelis, S. P., 2004.

1553: 	 Uses and abuses of fractal methodology in ecology.

1554: 	 {Ecol. Lett.} {7}, 254--271.,

1555: 	 doi:10.1111/j.1461-0248.2004.00568.x.

1556: %

1557:  \bibitem[Hara(1984)]{Hara84}

1558: 	 Hara, I., 1984.

1559: 	 Distribution and school size of Japanese sardine in the waters

1560: 	 off the southeastern coast of Hokkaido on the basis of echo

1561: 	 sounder surveys.

1562: 	 {Bull. Tokai Reg. Fish. Res. Lab.} {113}, 67--78.

1563: %

1564:  \bibitem[Hara(1986)]{Hara86}

1565: 	 Hara, I., 1986.

1566: 	 Stock assessment of Japanese sardine in the waters off the

1567: 	 southeast coast of Hokkaido using line transect method.

1568: 	 {Bull. Jpn. Soc. Sci. Fish.} {52}, 69--73.

1569: %

1570:  \bibitem[Hara(1990)]{Hara90}

1571: 	 Hara, I., 1990.

1572: 	 {Stock Assessment and Migration of Japanese Sardine

1573: 	 Schools in the Waters off the Southeastern Coast of Hokkaido}.

1574: 	 {Ph.D. Thesis}, Kyoto University, Kyoto, Japan.

1575: %

1576:  \bibitem[Harte {et al.}(1999)]{Harte-etal99}

1577: 	 Harte, J., Kinzig, A., Green, J., 1999.

1578: 	 Self-similarity in the distribution and abundance of species.

1579: 	 {Science} {284}, 334--336.

1580: %

1581:  \bibitem[Higashi and Yamamura(1993)]{Higashi-Yamamura93}

1582: 	 Higashi, M., Yamamura, N., 1993.

1583: 	 What determines animal group size? insider-outsider conflict

1584: 	 and its resolution.

1585: 	 {Am. Nat.} {142}, 553--563.

1586: %

1587:  \bibitem[Hoare {et al.}(2004)]{Hoare-etal04}

1588: 	 Hoare, D. J., Couzin, I. D., Godin, J.-G. J., Krause, J., 2004.

1589: 	 Context-dependent group size choice in fish.

1590: 	 {Anim. Behav.} {67}, 155--164.,

1591: 	 doi:10.1016/j.anbehav.2003.04.004.

1592: %

1593:  \bibitem[Hori(1977)]{hori77}

1594: 	 Hori, J., 1977.

1595: 	 {Langevin Equation}.

1596: 	 Iwanami Shoten, Publ., Tokyo.

1597: %

1598:  \bibitem[Keitt and Stanley(1998)]{Keitt-Stanley98}

1599: 	 Keitt, T. H., Stanley, H. E., 1998.

1600: 	 Dynamics of North American breeding bird populations.

1601: 	 {Nature} 393, 257--260.

1602: %

1603:  \bibitem[Keitt {et al.}(2002)]{Keitt-etal02}

1604: 	 Keitt, T. H., Amaral, L. A. N., Buldyrev, S. V., Stanley, H. E.,

1605: 	 2002.

1606: 	 Scaling in the growth of geographically subdivided populations:

1607: 	 invariant patterns from a continent-wide biological survey.

1608: 	 {Phil. Trans. R. Soc. Lond. B} {357}, 627--633.,

1609: 	 doi:10.1098/rstb.2001.1013.

1610: %

1611:  \bibitem[Kubo(1966)]{kubo66}

1612: 	 Kubo, R., 1966.

1613: 	 The fluctuation-dissipation theorem.

1614: 	 {Rep. Prog. Phys.} {29} (1), 255--284.

1615: %

1616:  \bibitem[Lillo {et al.}(2002)]{Lillo-etal02}

1617: 	 Lillo, F., Farmer, J. D., Mantegna, R. N., 2002.

1618: 	 Single curve collapse of the price impact function for the New

1619: 	 York Stock Exchange.

1620: 	 preprint cond-mat/0207428 at

1621: 			 $

1622: 			 \langle

1623: 			 \mbox{http://xxx.lanl.gov}

1624: 			 \rangle

1625: 			 $

1626: %

1627:  \bibitem[Lillo {et al.}(2003)]{Lillo-etal03}

1628: 	 Lillo, F., Farmer, J. D., Mantegna, R. N., 2003.

1629: 	 Master curve for price-impact function.

1630: 	 {Nature} {421}, 129--130.

1631: %

1632:  \bibitem[MacArthur and Wilson(1967)]{MacArthur-Wilson67}

1633: 	 MacArthur, R. H., Wilson, E. O., 1967.

1634: 	 {The Theory of Island Biogeography}.

1635: 	 Princeton University Press, Princeton.

1636: %

1637:  \bibitem[Mandelbrot(1982)]{Mandelbrot82}

1638: 	 Mandelbrot, B. B., 1982.

1639: 	 {The Fractal Geometry of Nature}.

1640: 	 W. H. Freeman and Co., New York.

1641: %

1642:  \bibitem[May(1999)]{May99}

1643: 	 May, R. M., 1999.

1644: 	 Unanswered questions in ecology.

1645: 	 {Phil. Trans. R. Soc. Lond. B} {354}, 1951--1959.

1646: %

1647:  \bibitem[Mayer and Mayer(1940)]{Mayer-Mayer40}

1648: 	 Mayer, M. G., Mayer, J. E., 1940.

1649: 	 {Statistical Mechanics}.

1650: 	 John Wiley \& Sons, New York.

1651: %

1652:  \bibitem[Misund(1990)]{Misund90}

1653: 	 Misund, O. A., 1990.

1654: 	 Sonar observations of schooling herring: school dimensions,

1655: 	 swimming behaviour, and avoidance of vessel and purse seine.

1656: 	 {Rapp. P.-v R\'eun. Cons. int. Explor. Mer} {189}, 135--146.

1657: %

1658:  \bibitem[Misund(1993a)]{misund93a}

1659: 	 Misund, O. A., 1993a.

1660: 	 Dynamics of moving masses: variability in packing density

1661:  shape and size among herring sprat and saithe schools.

1662: 	 {ICES J. Mar. Sci.} {50}, 145--160.

1663: %

1664: \bibitem[Misund(1993b)]{Misund93b}

1665: 	 Misund, O. A., 1993b.

1666: 	 Abundance estimation of fish schools based on a relationship

1667: 	 between school area and school biomass.

1668: 	 {Aquat. Living Resour.} {6}, 235--241.

1669: %

1670:  \bibitem[Misund {et al.}(1992)]{Misund-etal92}

1671: 	 Misund, O. A., Aglen, A., Beltestad, A. K., Dalen, J., 1992.

1672: 	 Relationship between the geometric dimensions and biomass of

1673: 	 schools.

1674: 	 {ICES J. Mar. Sci.} {49}, 305--315.

1675: %

1676: \bibitem[Misund {et al.}(1995)]{Misund-etal95}

1677: 	 Misund, O. A., Aglen, A., Fr{\o}n{\ae}s, E., 1995.

1678: 	 Mapping the shape, size, and density of fish schools by echo

1679: 	 integration and a high-resolution sonar.

1680: 	 {ICES J. Mar. Sci.} {52}, 11--20.

1681: %

1682:  \bibitem[Misund {et al.}(1996)]{Misund-etal96}

1683: 	 Misund, O. A., Aglen, A., Hamre, J., Ona, E., R{\o}ttingen, I.,

1684: 	 Skagen, D., Valdemarsen, J. W., 1996.

1685: 	 Improved mapping of schooling fish near the surface: comparison

1686: 	 of abundance estimates obtained by sonar and echo integration.

1687: 	 {ICES J. Mar. Sci.} {53}, 383--388.

1688: %

1689: \bibitem[Misund and Coetzee(2000)]{Misund-Coetzee00}

1690: 	 Misund, O. A., Coetzee, J., 2000.

1691: 	 Recording fish schools by multi-beam sonar: potential for

1692: 	 validating and supplementing echo integration recordings of

1693: 	 schooling fish.

1694: 	 {Fish. Res.} {47}, 149--159.,

1695: 	 doi:10.1016/S0165-7836(00)00166-1.

1696: %

1697: \bibitem[Misund {et al.}(2003)]{Misund-etal03}

1698: 	 Misund, O. A., Coetzee, J., Fr{\'e}on, P., Gardener, M., Olsen,

1699: 	 K., Svellingen, I., Hampton, I., 2003.

1700: 	 Schooling behaviour of sardine {\it Sardinops sagax} in False

1701: 	 Bay, South Africa.

1702: 	 {Afr. J. Mar. Sci.} {25}, 185--193.

1703: %

1704: \bibitem[Misund and Floen(1993)]{Misund-Floen93}

1705: 	Misund, O. A., Floen, S., 1993.

1706: 	Packing density structure of herring schools.

1707: 	{ICES Mar. Sci. Symp.} {196}, 26--29.

1708: %

1709:  \bibitem[Mogilner {et al.}(2003)]{Mogilner-etal03}

1710: 	 Mogilner, A., Edelstein-Keshet, L., Bent, L., Spiros, A., 2003.

1711: 	 Mutual interactions, potentials, and individual distance in a

1712: 	 social aggregation.

1713: 	 {J. Math. Biol.} {47}, 353--389.,

1714: 	 doi:10.1007/s00285-003-0209-7.

1715: %

1716:  \bibitem[Morgan(1988)]{Morgan88}

1717: 	 Morgan, M. J., 1988.

1718: 	 The effect of hunger, shoal size and the presence of a predator

1719: 	 on shoal cohesiveness in bluntnose minnows, {\it Pimephales

1720: 	 notatus} Rafinesque.

1721: 	 {J. Fish. Biol.} {32}, 963--971.

1722: %

1723:  \bibitem[Niwa(1994)]{Niwa94}

1724: 	 Niwa, H.-S., 1994.

1725: 	 Self-organizing dynamic model of fish schooling.

1726: 	 {J. Theor. Biol.} {171}, 123--136.

1727: %

1728:  \bibitem[Niwa(1996a)]{Niwa96a}

1729: 	 Niwa, H.-S., 1996a.

1730: 	 Newtonian dynamical approach to fish schooling.

1731: 	 {J. Theor. Biol.} {181}, 47--63.

1732: %

1733:  \bibitem[Niwa(1996b)]{Niwa96b}

1734: 	 Niwa, H.-S., 1996b.

1735: 	 Mathematical model for the size distribution of fish schools.

1736: 	 {Computers Math. Applic.} {32} (11), 79--88.

1737: %

1738:  \bibitem[Niwa(1998)]{Niwa98}

1739: 	 Niwa, H.-S., 1998.

1740: 	 School size statistics of fish.

1741: 	 {J. Theor. Biol.} {195}, 351--361.

1742: %

1743:  \bibitem[Niwa(2003)]{Niwa03}

1744: 	 Niwa, H.-S., 2003.

1745: 	 Power-law versus exponential distributions of animal group

1746: 	 sizes.

1747: 	 {J. Theor. Biol.} {224}, 451--457.,

1748: 	 doi:10.1016/S0022-5193(03)00192-9.

1749: %

1750:  \bibitem[Niwa(2004a)]{Niwa04a}

1751: 	 Niwa, H.-S., 2004a.

1752: 	 Space-irrelevant scaling law for fish school sizes.

1753: 	 {J. Theor. Biol.} {228}, 347--357.,

1754: 	 doi:10.1016/j.jtbi.2004.01.011.

1755: %

1756:  \bibitem[Niwa(2004b)]{Niwa04b}

1757: 	 Niwa, H.-S., 2004b.

1758: 	 Exploitation dynamics of fish stocks.

1759: 	 {Meeting Abstracts of the Physical Society of Japan}

1760: 	 59 (2), Part 2, p.201.

1761: %

1762: \bibitem[Okubo and Chiang(1974)]{Okubo-Chiang74}

1763: 	Okubo, A., Chiang, H. C., 1974.

1764: 	An analysis of the kinematics of swarming of {\it Anarete

1765: 	pritchardi} Kim (Diptera: Cecidomyiidae).

1766: 	{Res. Popul. Ecol.} {16}, 1--42.

1767: %

1768:  \bibitem[Okubo and Levin(2001)]{okubo-levin01}

1769: 	 Okubo, A., Levin, S. A., 2001.

1770: 	 {Diffusion and Ecological Problem: Modern Perspectives. 2nd ed.}

1771: 	 Springer-Verlag, New York.

1772: %

1773: \bibitem[Parr(1927)]{Parr27}

1774: 	Parr, A. E., 1927.

1775: 	A contribution to the theoretical analysis of the schooling

1776: 	behavior of fishes.

1777: 	{Bulletin of the Bingham Oceanographic Collection} {1} (1),

1778: 	1--32.

1779: %

1780:  \bibitem[Parrish and Hamner {(Eds.)}(1997)]{Parrish-Hamner97}

1781: 	 Parrish, J. K., Hamner, W. M. (Eds.), 1997.

1782: 	 {Animal Groups in Three Dimensions}.

1783: 	 Cambridge University Press, Cambridge.

1784: %

1785:  \bibitem[Partridge(1980)]{Partridge80}

1786: 	 Partridge, B. L., 1980.

1787: 	 The effect of school size or the structure and dynamics of

1788: 	 minnow schools.

1789: 	 {Anim. Behav.} {28}, 68--77.

1790: %

1791:  \bibitem[Partridge(1981)]{Partridge81}

1792: 	 Partridge, B. L., 1981.

1793: 	 Internal dynamics and interrelations of fish in schools.

1794: 	 {J. Comp. Physiol.} {144}, 313--325.

1795: %

1796:  \bibitem[Partridge(1982)]{Partridge82}

1797: 	 Partridge, B. L., 1982.

1798: 	 The structure and function of fish schools.

1799: 	 {Sci. Am.} {246}, 90--99.

1800: %

1801:  \bibitem[Pitcher(1973)]{Pitcher73}

1802: 	 Pitcher, T. J., 1973.

1803: 	 The three-dimensional structure of fish schools in the minnow,

1804: 	 {\it Phcxinus phcxinus}.

1805: 	 {Anim. Behav.} {21}, 673--686.

1806: %

1807:  \bibitem[Pitcher and Parrish(1993)]{Pitcher-Parrish93}

1808: 	 Pitcher, T. J., Parrish, J. K., 1993.

1809: 	 Functions of shoaling behaviour in teleosts,

1810: 	 in: Pitcher, T. J. (Ed.),

1811: 	 {Behaviour of Teleost Fishes. 2nd ed.}

1812: 	 Chapman \& Hall, London,

1813: 	 Ch.12, pp.363--439.

1814: %

1815:  \bibitem[Pitcher and Partridge(1979)]{Pitcher-Partridge79}

1816: 	 Pitcher, T. J., Partridge, B. L., 1979.

1817: 	 Fish school density and volume.

1818: 	 {Mar. Biol.} {54}, 383--394.

1819: %

1820:  \bibitem[Preston(1962)]{Preston62}

1821: 	 Preston, F. W., 1962.

1822: 	 The canonical distribution of commonness and rarity:

1823: 	 Part I, and Part II.

1824: 	 {Ecology} {43}, 185--215, and 410--432.

1825: %

1826:  \bibitem[Ptitsyn {et al.}(1968)]{Ptitsyn-etal68}

1827: 	 Ptitsyn, O. B., Kron, A. K., Eizner, Y. Y., 1968.

1828: 	 The models of the denaturation of globular proteins. I. theory

1829: 	 of globula-coil transitions in macromolecules.

1830: 	 {J. Polym. Sci. C} {16}, 3509--3517.

1831: %

1832:  \bibitem[Pulliam and Caraco(1984)]{Pulliam-Caraco84}

1833: 	 Pulliam, H. R., Caraco, T., 1984.

1834: 	 Living in groups: is there an optimal group size?,

1835: 	 in: Krebs, J.R., Davies, N.B. (Eds.),

1836: 	 {Behavioural Ecology. An Evolutionary Approach. 2nd ed.}

1837: 	 Blackwell Science, Oxford,

1838: 	 pp. 122--147.

1839: %

1840:  \bibitem[Rinaldo {et al.}(2002)]{Rinaldo-etal02}

1841: 	 Rinaldo, A., Maritan, A., Cavender-Bares, K. K.,

1842: 	 Chisholm, S. W., 2002.

1843: 	 Cross-scale ecological dynamics and microbial size spectra in

1844: 	 marine ecosystems.

1845: 	 {Proc. R. Soc. Lond. B} {269}, 2051--2059.,

1846: 	 doi: 10.1098/rspb.2002.2102.

1847: %

1848:  \bibitem[Rouse(1953)]{Rouse53}

1849: 	 Rouse, P. E., 1953.

1850: 	 A theory of the linear viscoelastic properties of dilute

1851: 	 solutions of coiling polymers.

1852: 	 {J. Chem. Phys.} {21}, 1272--1280.

1853: %

1854:  \bibitem[Stanley(1995)]{Stanley95}

1855: 	 Stanley, H. E., 1995.

1856: 	 Power laws and universality.

1857: 	 {Nature} {378}, 554.

1858: %

1859: \bibitem[Symons(1971)]{Symons71}

1860: 	Symons, P. E. K., 1971.

1861: 	Spacing and density in schooling threespine sticklebacks

1862: 	({\it Gasterosteus aculeatus}) and mummichog ({\it Fundulus

1863: 	heteroclitus}).

1864: 	{J. Fish. Res. Bd. Can.} {28}, 999--1004.

1865: %

1866:  \bibitem[Tauti and Miyosi(1929)]{Tauti-Miyosi29}

1867: 	 Tauti, M., Miyosi, K., 1929.

1868: 	 On the movement of a fish-group. I.

1869: 	 on the direction of a fish-group on passing the end of a wall

1870: 	 along which it has moved.

1871: 	 {J. Imperial Fish. Inst. Jpn.} {24}, 73--76.

1872: %

1873:  \bibitem[Tauti and Hudino(1929)]{Tauti-Hudino29}

1874: 	 Tauti, M., Hudino, T., 1929.

1875: 	 On the movement of a fish-group. II.

1876: 	 the case when fishes were barred from their progress by a wall.

1877: 	 {J. Imperial Fish. Inst. Jpn.} {24}, 76--79.

1878: %

1879:  \bibitem[Tauti and Yasuda(1929)]{Tauti-Yasuda29}

1880: 	 Tauti, M., Yasuda, H., 1929.

1881: 	 On the movement of a fish-group. III.

1882: 	 the case when fishes were barred from their progress by a net.

1883: 	 {J. Imperial Fish. Inst. Jpn.} {24}, 80--86.

1884: %

1885:  \bibitem[Tauti and Yasuda(1930)]{Tauti-Yasuda30}

1886: 	 Tauti, M., Yasuda, H., 1930.

1887: 	 On the movement of a fish-group. IV.

1888: 	 the shape of the entrance of fishing net and the fish-group

1889: 	 moving in and out.

1890: 	 {J. Imperial Fish. Inst. Jpn.} {25}, 41--50.

1891: %

1892:  \bibitem[Tauti and Yasuda(1933a)]{Tauti-Yasuda33a}

1893: 	 Tauti, M., Yasuda, H., 1933a.

1894: 	 The direction in which a fish-group goes on after having met

1895: 	 with a net.

1896: 	 {Bull. Jpn. Soc. Sci. Fish.} {2}, 17--19.

1897: %

1898:  \bibitem[Tauti and Yasuda(1933b)]{Tauti-Yasuda33b}

1899: 	 Tauti, M., Yasuda, H., 1933b.

1900: 	 How will a fish-group be attracted?

1901: 	 {Bull. Jpn. Soc. Sci. Fish.} {2}, 124--127.

1902: %

1903:  \bibitem[Uhlenbeck and Ford(1962)]{Uhlenbeck-Ford62}

1904: 	 Uhlenbeck, G. E., Ford, G. W., 1962.

1905: 	 The theory of linear graphs with applications to the theory of

1906: 	 the virial development of the properties of gases,

1907: 	 in: de Boer, J., Uhlenbeck, G. E. (Eds.),

1908: 	 {Studies in Statistical Mechanics}.

1909: 	 North-Holland, Amsterdam,

1910: 	 Vol.1, Ch.2, pp.119--211.

1911: %

1912:  \bibitem[Vicsek(2001)]{vicsek01}

1913: 	 Vicsek, T., 2001.

1914: 	 A question of scale.

1915: 	 {Nature} {411}, 421.

1916: %

1917:  \bibitem[Vicsek {(Ed.)}(2001)]{vicsek01-book}

1918: 	 Vicsek, T. (Ed.), 2001.

1919: 	 {Fluctuations and Scaling in Biology}.

1920: 	 Oxford University Press, Oxford.

1921: %

1922:  \bibitem[Voronoi(1908)]{voronoi08}

1923: 	 Voronoi, G. M., 1908.

1924: 	 {Nouvelles applications des param{\`e}tres continus {\`a} la

1925: 	 th{\'e}orie des formes quadratiques deuxi{\`e}me m{\'e}moire:

1926: 	 rechrches sur les parall{\'e}llo{\`e}dres primitifs}.

1927: 	 {J. Reine Angew. Math.} {134}, 198--287.

1928: %

1929:  \bibitem[Voronoi(1909)]{voronoi09}

1930: 	 Voronoi, G. M., 1909.

1931: 	 {Deuxi{\`e}me m{\'e}moire: recherches sur les

1932: 	 parall{\'e}loedres primitifs}.

1933: 	 {J. Reine Angew. Math.} {136}, 67--181.

1934: %

1935:  \bibitem[Warburton and Lazarus(1991)]{Warburton-Lazarus91}

1936: 	 Warburton, K., Lazarus, J., 1991.

1937: 	 Tendency-distance models of social cohesion in animal groups.

1938: 	 {J. Theor. Biol.} {150}, 473--488.

1939: %

1940:  \bibitem[Wilson(1983)]{wilson83}

1941: 	 Wilson, K. G., 1983.

1942: 	 The renormalization group and the critical phenomena.

1943: 	 {Rev. Mod. Phys.} {55}, 583-600.

1944: %

1945: \end{thebibliography}

1946: %

1947: \end{document}

1948: