ICES Journal of Marine Science: Journal du Conseil Advance Access originally published online on June 3, 2008
ICES Journal of Marine Science: Journal du Conseil 2008 65(7):1144-1152; doi:10.1093/icesjms/fsn090
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The use of sagittal otoliths in discriminating stocks of common dolphinfish (Coryphaena hippurus) off northeastern Brazil using multishape descriptors
1 Unidade Acadêmica de Garanhuns, Universidade Federal Rural de Pernambuco, Rua Bom Pastor s/n, Mundau, Garanhuns – PE CEP 55296-901, Brazil
2 Laboratório de Dinâmica de Populações Marinhas, Departamento de Pesca, Universidade Federal Rural de Pernambuco, Avenida D. Manoel de Medeiros s/n, Dois Irmãos, Recife – PE CEP 52171-900, Brazil
3 Departamento de Estatística e Informática, Universidade Federal Rural de Pernambuco, Avenida D. Manoel de Medeiros s/n, Dois Irmãos, Recife – PE CEP 52171-900, Brazil
4 Laboratoire de Sclérochronologie des Animaux Aquatiques, Institut de Recherche pour le Développement, Centre IRD de Bretagne, BP 70, 29280 Plouzané, France
Correspondence to P. Duarte-Neto: tel: +55 87 37610969; fax: +55 87 37610882; e-mail: pjneto{at}uag.ufrpe.br
Duarte-Neto, P., Lessa, R., Stosic, B., and Morize, E. 2008. The use of sagittal otoliths in discriminating stocks of common dolphinfish (Coryphaena hippurus) off northeastern Brazil using multishape descriptors. – ICES Journal of Marine Science, 65: 1144–1152.The shape of sagittal otoliths from the dolphinfish (Coryphaena hippurus) was studied to test the hypothesis that two stocks exist off the northeast coast of Brazil. In all, 82 sagittal otoliths were collected from fish caught by the artisanal fleet in two coastal regions of northeastern Brazil in December 2003 and April/May 2004. Several shape descriptors were determined [area, perimeter, rectangularity, circularity, eccentricity, fractal dimension (FD), and Fourier coefficients (FCs)] to evaluate the degree of similarity in the otoliths between regions. A three-morphotype pattern was revealed through cluster and principal component analyses based on FCs of the 30th harmonics. Apparently, this pattern is not influenced by clinal factors. Despite the great variability between otolith shapes, separation of the samples from two regions was suggested using multivariate and univariate analyses of variance for all shape descriptors and using canonical discriminant analysis. The methods successfully classified 57.1 and 69.6% of otoliths from the Maranhão and Rio Grande do Norte regions, respectively. The FD was a powerful descriptor in discriminating the two stocks. Differences in the shapes of sagittal otoliths may be related to different growth rates of the species and lend credence to the belief that there are two stocks along the Brazilian coast.
Keywords: Coryphaena hippurus, fish stock, multivariate analysis, otolith shape
Received 13 April 2007; accepted 10 May 2008; advance access publication 3 June 2008.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionReferences |
|---|
The common dolphinfish (Coryphaena hippurus) is an epipelagic oceanic fish, distributed worldwide in tropical and subtropical waters (Gibbs and Collette, 1959; Johnson, 1978) and is usually restricted to water warmer than 20°C (Gibbs and Collette, 1959). It is an important resource, supporting commercial and sport fisheries throughout its range (Palko et al., 1982). In the area 30–40°W 0–12°N, it accounts for 12% of artisanal landings (IBAMA, 2000), represents 7.3% of catches in fisheries targeting swordfish (Evangelista et al., 1998), and 18% of captures in surveys of large pelagic fish between 1992 and 1997 (Lessa, 2003).
Based on catch seasonality, mean sizes in catches, life history, and genetic characteristics, Oxenford and Hunte (1986) proposed two unit stocks with different migratory circuits in the Caribbean. The same authors suggested a northeastern migratory circuit incorporating the northern Caribbean islands, the southeastern US, and Bermuda, and a southeastern migratory circuit incorporating the southeastern Caribbean Islands and the north and northeastern Brazilian coast. From that hypothesis of migration, abundance could peak off northeastern Brazil between June and January. Considering the distance between the Virgin Islands and the northeast coast of Brazil, high abundance off NE Brazil from September to October would fit the migration circuit proposed. As catch data can be used as crude indicators of stock structure (Begg and Waldman, 1999), Lessa (2003) suggested another migratory circuit restricted to the east coast of Brazil, also based on fishery seasonality (peaks of abundance) and mean size in the catches of dolphinfish (Figure 1). She found different periods of high catch rates, which did not fit the Oxenford and Hunte (1986) hypothesis for the waters off Brazil, and instead proposed the existence of an eastern Brazilian stock migratory circuit, running from the state of Rio Grande do Norte (RN) down to the south of the country.
|
Differences in size composition between areas can be indicative of different degrees of exploitation, perhaps due to selectivity factors, or perhaps differences in the life history of the Brazilian and Caribbean stocks. However, the two different putative stocks along the Brazilian coast, defined by size (Lessa, 2003), raises a set of hypotheses to be tested. Moreover, in a study of age and growth of dolphinfish caught off NE Brazil, Lessa et al. (2004) observed considerable variation in the shape of sagittal otoliths. However, such dissimilarity could not be clearly related to any particular factor and remains unexplained. Therefore, an analysis of otolith shape was considered critical to testing the hypothesis of there being two Brazilian stocks. It is necessary to determine whether the variations in shape are attributable to the existence of two stocks in the area or to individual fish variability.
Otolith shape is highly species-specific (Gaemers, 1984; Nolf, 1985), and frequently exhibits clinal variation related to distance, geographic location (Worthman, 1979), and depth (Wilson, 1985). Gauldie (1988) reported how otolith shape may be controlled by several factors, including checks and discontinuities in growth controlled by the sensory epithelium. Consequently, these stimuli are induced by variations in environmental conditions, such as photoperiod, temperature, and feeding habits (Lecomte-Finiger, 1999). These characteristics make the otolith an ideal structure for stock identification, because it contains a series of characteristic measures, including optical density, micro-elements, microstructural zonation, and growth patterns, as well as linear and shape morphometry (Campana and Neilson, 1985). Morphometry has been used in stock identification of marine fish such as Atlantic mackerel (Scomber scombrus; Castonguay et al., 1991), silver hake (Merluccius bilinearis; Bolles and Begg, 2000), haddock (Melanogrammus aeglefinus; Begg et al., 2000), and king mackerel (Scomberomorus cavalla; De Vries et al., 2002). Image analysis, such as of otolith shape, plays a significant role in morphometric research, improving data collection and providing a better description of shape and the potential for developing new analytical methods (Cadrin and Friedland, 1999).
The aim of our study was to analyse the shape of sagittal otoliths in the common dolphinfish to explain the variability in otolith shape, allowing the hypothesis of there being two separate stocks in the study area to be tested.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Sampling and image acquisition
Samples were collected in December 2003 off the State of Maranhão (MA), supposedly representative of the southern Caribbean circuit, and in April/May 2004 off the State of RN, supposedly representative of the eastern Brazil circuit; both States are located along the northeast coast of Brazil (Figure 1). Fish in the sample were limited to 0.5–1.0 years old to minimize age-related variations that could confound discrimination analyses. Absolute ages in years were estimated using the growth parameters determined by Lessa et al. (2004).
Fish were measured (fork length, FL, cm), and the sex was determined by the external sexual dimorphism of the forehead (Palko et al., 1982). In all, 82 pairs of sagittal otoliths (35 from MA, and 47 from RN) were removed, cleaned, and stored dry.
The right and left otoliths were orientated horizontally with the sulcus acusticus upwards and the rostrum and antirostrum to the left (Figure 2). Otolith images were captured using a charge-coupled device camera mounted on a microscope. Image processing was carried out using an image-analysis system developed for calcified structures (TNPC: Visilog software platform, NOESIS, France).
|
Shape descriptors
A number of shape descriptors was determined to evaluate the degree of similarity in simple and complex otolith morphological characteristics between the two areas. Two morphometric shape descriptors (perimeter, P; and area, A), otolith length, and three shape indices (rectangularity, circularity, and eccentricity; Table 1) were obtained using TNPC. Rectangularity (R) describes the length and width variations with respect to area, where 1.0 represents a perfect rectangle. Circularity (CI) measures the similarity of a structure to a perfect circle, using a minimum value of 4
(12.57). Eccentricity (E) describes the variability of the configuration of contour points in relation to the centre of mass, calculated using the second-order moments of the contour (Russ, 1990).
|
The fractal dimension (FD) of the otolith contour was also calculated, because fractal measurements are widely applied in different problems concerning image-processing analysis and pattern recognition (Costa and Cesar, 2001). Fractal analysis does not allow contour comparison regarding shape, but rather with regard to complexity. Therefore, a higher FD indicates a greater complexity of the structure. A software program was constructed in the C language to determine the FD, using the box-counting method (Bassingthwaighte et al., 1994; Pierra et al., 2005).
Fourier coefficients (FCs) were calculated using elliptical Fourier analysis (EFA) on the SHAPE software program, version 1.2 (Iwata, 2002), which allows more efficient calculation of the coefficients with regard to the flexibility and consistency of results when compared with other Fourier analyses (Rohlf and Archie, 1984). The method decomposes the contour through variations in the x and y coordinates separately (
x and y), as a parametric function of distance in the cumulative chord (t) along the contour (Kuhl and Giardina, 1982). The distance is proportioned always to vary from 0 to 2 along the contour. The two parametric functions are defined in x(t) and y(t), respectively, as
|
|
|
|
|
|
|
|
|
|
|
|
Statistical analyses
Morphometric variables were first examined for normality and homogeneity of variance, then log-transformed before statistical analysis if these criteria were not satisfied. The effect of otolith length on the magnitude of each otolith variable was determined using analysis of covariance (ANCOVA), except for the FCs, which were interpreted as invariant with respect to size (Diaz et al., 1997). Variables that were significantly correlated with otolith length were corrected for variable otolith length using the common within-group slope b (Lleonart et al., 2000).
Multivariate analysis of variance (MANOVA) was used to compare the symmetry between the right and the left otolith, to determine differences between sexes, and to determine whether these factors have an influence on the variation in otolith shape.
To assess a shape pattern among otoliths without prior definition by region, two multivariate techniques were employed, using the FC data matrix for the 30th harmonics. The first was cluster analysis (CA) based on the Ward method (Reis, 1997); the Euclidian distance was the similarity index among otoliths. The second was principal components analysis (PCA), which was also applied to reduce the dimensionality of the data. The significant principal components (PCs) were established according to the Kaiser criterion (Reis, 1997). The PC scores were used in the multivariate analysis between regions.
MANOVA was used to compare the dolphinfish otolith samples between the two regions. Univariate analysis of variance (ANOVA) was then used to examine individual variables to explain any significant differences detected through MANOVA.
The class separation distance (CSD) was used to determine which shape descriptors maximized the class separation (Costa and Cesar, 2001), evaluating the potential performance of the classification based on a single feature (Pierra et al., 2005). The CSD between 
and β classes with respect to the mth descriptor is defined as
|
|
2 represent the mean and variance of the mth feature computed over the samples associated with the and β classes. The classification potential increases with CSD, β, m. Canonical discriminant analysis (CDA) was then used to detect morphometric differences in the otoliths of dolphinfish from the two regions. Significant (p < 0.05) canonical variables provided the best overall discrimination between groups. Standardized coefficients provided the contribution of each shape descriptor towards the discrimination of samples between regions. Jackknifed cross-validation procedures were used to generate an unbiased classification matrix.
All statistical analyses were performed on the MINITAB version 13.2 (MINITAB Inc.) and STATISTICA version 6.0 (StatSoft Inc.) software programs, except the PCA, which was performed on SHAPE version 1.2.
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Only CI and P variables were log-transformed to correct for non-normality and heterogeneity of variances. ANCOVA indicated that E, R, and FD were not correlated with otolith length (p > 0.05). The other descriptors were corrected for variable fish length using their respective common within-group slope (Table 2).
|
No significant differences were found (MANOVA, p > 0.05) between right and left otoliths, so in cases where the right otolith was missing, the left was used. Comparison of the otoliths between sexes indicated no significant differences (MANOVA, p > 0.05), so all analyses were performed with sexes combined.
General shape analysis
Tree clustering was performed from the amplitudes of the 30th harmonics, with similar otoliths grouped (Figure 3). Three general groups of shapes were established with up to 50% of the maximum Euclidian distance analysed. The three morphotypes are displayed in Figure 3. Morphotype I has dorsal and ventral wings of similar size, and Morphotypes II and III have much smaller dorsal wings than ventral wings. III differs from II in the lack of separation of the wings posteriorly.
|
In total, 12 PCs were considered to be significant. However, there was stability in the eigenvalues starting from PC6 (Figure 4), so only the first six components were used, explaining 83.4% of the total variance. When the shape variation (mean ± 2 s.d.; i.e. the square root of the eigenvalue of the component) explained by each component (Figure 5) is visualized, the principal characteristics exhibited by the components are the size of the dorsal wing in relation to the ventral wing, and the posterior connection between the two wings.
|
|
PC1 vs. PC3 exhibits the same groups as the CA, whereas for the others, only Morphotype III was prone to identification (Figure 6). At least three morphotypes of dolphinfish otoliths were evident, because the first and third components are strongly related to the characteristics above. As PC1 retained a major portion of the total variance explained, its contribution was nearly homogeneous for most of the harmonics, but with a slight increase in the final harmonics, showing their importance in the otolith contour of dolphinfish (Figure 7). PC2 and PC3 were more related to the first harmonics, and PC4 more related to the first and last harmonics. PC5 and PC6 presented the same tendency as PC4, but with a lesser number of correlated harmonics.
|
|
ANOVA on the significant PCs among the three morphotypes (Table 3) indicates that PC1, PC2, and PC3 are related to variations between shape groups, and PC4 is related to variations between geographic locations, according to the comparison between regions. No significant differences were found for the last two components in either comparison. Therefore, the harmonics correlated with PC4 contained greater shape variations between stocks, because these harmonics corresponded to those that had either a slight correlation or no correlation with the third components (Figure 7d).
|
ANOVA was used to evaluate the correlation between morphotypes and fish age, grouping the absolute ages of fish by morphotype; no significant differences were found (p > 0.05).
Analysis between regions
Significant multivariate differences were detected between regions (MANOVA, p = 0.0137). ANOVA indicated that E, FD, and PC4 were the significant variables (Table 4). Low CSD values for most PCs again showed that the otolith contour variability of dolphinfish was more related to individual differences and less to stock differences.
|
Discriminant analyses provided further support for the separation of the two regions studied, despite the high variability among otolith shapes suggesting two stocks of dolphinfish along the northeastern Brazilian coast, as indicated by the multivariate and univariate analyses (Figure 8). Significant differences were detected between the scores of the regions (ANOVA, p < 0.001). Overall, 57.1% of the MA sample and 69.6% of the RN sample were successfully classified. Variations in R, A, and FD accounted for the separation of the two samples (Table 5).
|
|
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
According to Cadrin and Friedland (1999), Fourier analysis is an efficient method for describing contour shapes. However, understanding the reason for subtle differences in shape is abstract. Here, this difficulty was minimized by a combined application of PCA and CA to interpret the FCs, demonstrating that these methods have utility in this type of analysis.
Authors such as Campana and Casselman (1993) used harmonics as a PC. However, unlike PCs, harmonics generally do not explain variance in decreasing order (Kaesler, 1997), so leading to the exclusion of some suitable harmonics. For example, the slight increase in the last harmonics presented in the PC1 contribution (Figure 7a) demonstrates the importance of these harmonics in the variation in dolphinfish otolith shape. Therefore, PCA permitted a reduction in dimensionality regarding the effect of all harmonics on the shape of the otolith.
Three morphotypes were determined based on the contour of dolphinfish otoliths (Figure 4). Ehrlich et al. (1983) and Bird et al. (1986) demonstrated that coefficients from the first harmonics can be interpreted and related to general otolith shape. Therefore, we can conclude that the characteristics of each morphotype are related to general differences in the otolith shape of dolphinfish, because they are more correlated with the first harmonic (Figure 7).
Stock differences have little influence on the determination of morphotypes, as shown in Table 3, because significant differences were only found for PC4. Therefore, as the results do not indicate separation by geography, there is no evidence of clinal balancing selection. We speculate that the variation between morphotypes may be maintained by temporal balancing selection. According to Gauldie and Crampton (2002), one type is maintained in 1 year and another type the next year; the first type is then repeated again in the following year, etc., following this basic pattern. The effect is that, at any given time, if the selection of a unique type is not absolute, there is a persistent polymorphism.
In comparing MA and RN samples, differences were found in the structure of the sagittal otolith of dolphinfish in both the MANOVA and ANOVA as well as in CDA, providing a phenotypic base for stock separation along the Brazilian coast.
Among the descriptors used here, FD was the most efficient at discriminating the two samples, considering the significance always and high CSD values, and taking into account that it was not correlated with otolith length. Fractal analysis has been applied in several fields (behavioural, morphology, ecology, evolution, physiology, geography, physics, and chemistry; Monteiro and Reis, 1999). Apparently, however, there is no evidence of the utilization of these descriptors in stock discrimination using otoliths. Nevertheless, it is a useful tool that may be used in this type of study as well as in taxonomic studies based on otolith shape.
Surprisingly, the FCs represented by the PCs were the shape descriptors of dolphinfish otoliths with the least contribution to stock discrimination, usually. The third components showed significant differences between morphotypes (Table 3); PC4 was the only one that demonstrated differences between regions and a higher CSD value (Table 4). This indicates that most contour shape variability in the dolphinfish otolith is related to individual differences or occurred between morphotypes; the information corresponding to the regions was only contained in the harmonics correlated with PC4 (Figure 7d). This regional variation is reflected in both general and minor shape differences.
Therefore, we suggest that it is the small contribution of the FCs to stock separation, and consequently, the moderate success of the otolith classification between the two samples (57.1% for MA and 69.6% for RN) that may be related to high individual variability in otolith shape, because Fourier analysis is a method that provides contour representation with a high degree of precision (Rohlf and Archie, 1984). The other descriptors are more related to the basic characteristics of the otolith and less to the contour shape. As in Cardinale et al. (2004), this indicates that otolith classification is a difficult task requiring complex statistical tools and the combination of several descriptors.
In studying stock discrimination of cod (Gadus morhua), Campana and Casselman (1993) showed that otolith shape changes largely in response to differences in growth rate. Therefore, we suggest that the differences in dolphinfish otolith shape found here between the study regions are related to growth differences. Such differences are observed when comparing the von Bertalanffy growth coefficients (K) estimated for dolphinfish caught in their southeastern Caribbean circuit (K = 2.19, Rivera and Appeldoorn, 2000; K = 2.87, Oxenford and Hunte, 1983) with those estimated for fish caught in their eastern Brazil circuit (K = 0.897, Lessa et al., 2004), despite differences between the age determination methods employed. Consequently, the dolphinfish that constitute the southeastern Caribbean migratory circuit (including northern Brazil) would have faster growth rates than dolphinfish following the eastern Brazil circuit.
Life history (i.e. growth) and phenotypic (i.e. morphometric) interpretation of stock differences represent a significant advantage in using otolith shape over other stock identification methods, not to mention the logistical benefits (time and expense). In genetic studies, for example, recent divergence or substantial secondary reproductive contact results in no apparent differences in gene frequency between groups, even when such a difference exists (Begg and Waldman, 1999). Moreover, stock identification based on differences in life history has a clear biological interpretation and may be an indicator of interspecific grouping with unequal dynamics, which is central to stock assessment and management (Cadrin and Friedland, 1999; Cadrin, 2000).
Analysis of the sagittal otolith shape of dolphinfish in relation to differences in growth rate offers evidence for the hypothesis of two dolphinfish migratory circuits along the Brazilian coast, as proposed by Lessa (2003), even with the high individual variability. However, these differences do not provide a genetic basis for the separation. They may reflect the phenotypic characteristics of each stock, suggesting distinct life histories, so the differences in size composition observed by Lessa (2003) are probably related to independent recruitment or biological or fishery factors other than gear selectivity. Nevertheless, there is a lack of biological information for eastern Brazil circuit dolphinfish that could be helpful for understanding the differing dynamics of the two stocks. Moreover, evidence for the separation of the fish following these migratory circuits may be useful in defining dolphinfish stocks in the Atlantic and, consequently, may be used to underpin the development of appropriate management measures.
|
Acknowledgements |
|---|
Our study was funded by the Ministerio do Meio Ambiente—MMA, Secretaria da Comissão Interministerial para os Recursos do Mar—SECIRM, within the scope of "Programa Nacional de Avaliação do Potencial Sustentável dos Recursos Vivos—REVIZEE". The Conselho Nacional de Desenvolvimento Científico e Tecnológico—CNPq provided scholarships and research grants (Proc. 301048/83-Oc). The Brazilian authors benefited from a bilateral programme run by CNPq and France’s Institut de Recherche pour le Développement—IRD, through which facilities at the Laboratoire de Sclérochronologie des Animaux Aquatiques—LASAA, Brest, were made available. We also thank Netuno SA and Eng. Gustavo Oliveira for support. The experimental work complied with the current laws of Brazil. Finally, we are indebted to Ronan Fablet and André Ogor (IFREMER/LASAA) for encouragement and support, and two anonymous referees for valuable comments on the submitted draft.
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
-
Bassingthwaighte J. B., Liebovitch L. S., West B. J. Fractal Physiology. (1994) Oxford, UK: Oxford University Press.
Begg G. A., Overholtz W. J., Munroe N. J. The use of internal otolith morphometrics for identification of haddock (Melanogrammus aeglefinus) stocks on Georges Bank. Fishery Bulletin US (2000) 99:1–14.
Begg G. A., Waldman J. R. An holistic approach to fish stock identification. Fisheries Research (1999) 43:35–44.[CrossRef][Web of Science]
Bird J. L., Eppler D. T., Checkley D. M. Comparisons of herring otoliths using Fourier series shape analysis. Canadian Journal of Fisheries and Aquatic Sciences (1986) 43:1228–1234.
Bolles K. L., Begg G. A. Distinction between silver hake (Merluccius bilinearis) stocks in US waters of the Northwest Atlantic based on whole otolith morphometrics. Fishery Bulletin US (2000) 98:451–462.
Cadrin S. X. Advances in morphometric identification of fishery stocks. Reviews in Fish Biology and Fisheries (2000) 10:91–112.[CrossRef][Web of Science]
Cadrin S. X., Friedland K. D. The utility of image processing techniques for morphometrics analysis and stock identification. Fisheries Research (1999) 43:129–139.[CrossRef][Web of Science]
Campana S. E., Casselman J. M. Stock discrimination using otolith shape analysis. Canadian Journal of Fisheries and Aquatic Sciences (1993) 50:1062–1083.
Campana S. E., Neilson J. D. Microstructure of fish otoliths. Canadian Journal of Fisheries and Aquatic Sciences (1985) 42:1014–1032.
Cardinale M., Doering-Arjes P., Gangnon P. Effects of sex, stock, and environment on the shape of known-age Atlantic cod (Gadus morhua) otoliths. Canadian Journal of Fisheries and Aquatic Sciences (2004) 61:158–167.
Castonguay M., Simard P., Gagnon P. Usefulness of Fourier analysis of otolith shape for Atlantic mackerel (Scomber scombrus) stock discrimination. Canadian Journal of Fisheries and Aquatic Sciences (1991) 48:296–302.
Costa L. F., Cesar R. M. Shape Analysis and Classification: Theory and Practice. (2001) New York: CRC Press. 659.
De Vries D. A., Grimes C. B., Prager M. H. Using otolith shape analysis to distinguish eastern Gulf of Mexico and Atlantic Ocean stocks of king mackerel. Fisheries Research (2002) 57:51–62.[CrossRef][Web of Science]
Diaz G., Cappai C., Setzu M. D., Sirigu S., Diana A. Elliptic Fourier descriptors of cell and nuclear shapes. In: Fourier Descriptors and their Applications in Biology—Lestrel P. E., ed. (1997) Cambridge, UK: Cambridge University Press. 307–321. 465 pp.
Ehrlich R., Baxter Pharr R., Healy-Willians N. Comments on the validity of Fourier descriptors in systematics: reply to Bookstein et al. Systematic Zoology (1983) 32:202–206.
Evangelista J. E., Oliveira G. M., Vasconcelos J. A. A Evolução da Pesca no Nordeste do Brasil. Boletim Técnico-Científico do CEPENE (1998) 6:77–109.
Gaemers P. A. M. Taxonomic position of the Cichlidae (Pisces: Perciformes) as demonstrated by the morphology of their otoliths. Netherlands Journal of Zoology (1984) 34:566–595.
Gauldie R. W. Function, form and time-keeping properties of fish otoliths. Comparative Biochemistry and Physiology A (1988) 91:395–402.[CrossRef]
Gauldie R. W., Crampton J. S. An eco-morphological explanation of individual variability in the shape of the fish otolith: comparison of the otolith of Hoplostethus atlanticus with other species by depth. Journal of Fish Biology (2002) 60:1204–1221.[CrossRef][Web of Science]
Gibbs R. H., Collette B. B. On the identification, distribution, and biology of the dolphins, Coryphaena hippurus and C. equiselis. Bulletin of Marine Science of the Gulf of Caribbean (1959) 9:117–152.
IBAMA. Boletim Estatístico da Pesca Marítima e Estuarina (ESTATPESCA) do Nordeste do Brasil. In: Centro de Pesquisa e Extensão Pesqueira do Nordeste—CEPENE, Tamandaré, PE (2000) 135.
Iwata H. A software package for quantitative evaluation of biological shapes based on elliptic Fourier descriptors. Department of Information Science and Technology, (2002) Japan: National Agricultural Research Center.
Johnson G. D. Development of fishes of the Mid-Atlantic Bight. An atlas of egg, larval, and juvenile stages. Vol. 4. Carangidae through Ephippidae. US Department of the Interior, Fish and Wildlife Service, Biological Service Program, (1978) FWS/OBS-78/12:123–128.
Kaesler R. L. Phase angles, harmonic distance, and the analysis of form. In: Fourier Descriptors and their Applications in Biology—Lestrel P. E., ed. (1997) Cambridge, UK: Cambridge University Press. 106–125. 465 pp.
Kuhl F. P., Giardina C. R. Elliptic Fourier features of a closed contour. Computer Graphics and Image Processing (1982) 18:236–258.[CrossRef][Web of Science]
Lecomte-Finiger R. L’otolithe: la boîte des Téléostéens. Année de Biologie (1999) 38:107–122.
Lessa R. P. Dinâmica das frotas pesqueiras do Nordeste. Relatório REVIZEE/SCORE-NE (2003) vol. 1:108.
Lessa R. P., Santana F. M., Nogueira G. D. Idade e crecimento e avaliação do dourado, Coryphaena hippurus. In: Dinâmica de Populações e Avaliação de Estoques—Lessa R. P., Nobrega M. F., Bezerra J. L., eds. (2004) InRelatório REVIZEE/SCORE-NE. 27–38. 265 pp.
Lestrel P. E. Introduction and overview of Fourier descriptors. In: Fourier Descriptors and their Applications in Biology—Lestrel P. E., ed. (1997) Cambridge, UK: Cambridge University Press. 22–44. 465 pp.
Lleonart J., Salat J., Torres G. J. Removing allometric effects of body size in morphological analysis. Journal of Theoretical Biology (2000) 205:85–93.[CrossRef][Web of Science][Medline]
Monteiro L. R., Reis S. F. Princípios de Morfometria Geométrica. In: Holos, Ribeirão Preto (1999) 198.
Nolf D. Otolith piscium. In: Handbook of Paleoichthyology—Schultze H. P., ed. (1985) New York, NY: Gusttav Fisher Verlag. 1–45. 145 pp.
Oxenford H. A., Hunte W. Age and growth of dolphin, Coryphaena hippurus, determined by growth rings in otoliths. Fishery Bulletin US (1983) 84:906–909.
Oxenford H. A., Hunte W. A preliminary investigation of the stock structure of the dolphin, Coryphaena hippurus, in the western central Atlantic. Fishery Bulletin US (1986) 84:451–460.
Palko B. J., Beardsley G. L., Richards W. J. Synopsis of the biological data on dolphin-fishes, Coryphaena hippurus Linnaeus, and Coryphaena equiselis Linnaeus. In: FAO Fisheries Synopsis, 130 (1982) 28.
Pierra J., Parisi-Baradad V., Garcia-Ladona E., Lombarte A., Recasens L., Cabestany J. Otolith shape feature extraction oriented to automatic classification with open distributed data. Marine and Freshwater Research (2005) 56:805–814.[CrossRef][Web of Science]
Rivera G. A., Appeldoorn R. S. Age and growth of dolphinfish, Coryphaena hippurus, off Puerto Rico. Fishery Bulletin US (2000) 98:345–352.
Reis E. Estatística Multivariada Aplicada (1997) Lisboa: Edições Sílabo. 343.
Rohlf F. J., Archie J. W. A comparison of Fourier methods for description of wing shape in mosquitoes (Diptera: Culicidae). Systematic Zoology (1984) 33:302–317.
Russ J. C. Computer-Assisted Microscopy: the Measurement and Analysis of Images. (1990) New York: Plenum Press. 465.
Wilson R. R. Depth related changes in sagitta morphology in six macrourid fishes of the Pacific and Atlantic Oceans. Copeia (1985) 1985:1011–1017.[CrossRef]
Worthman H. A. Relação entre o desenvolvimento do otólito e o crescimento do peixe como auxílio na distinção de populações de pescada (Plagioscion squamosissimus). Acta Amazônica (1979) 9:573–586.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||