Considerações sôbre amostragem de peixes marinhos (II)

Nomura, H.

doi:10.1590/S0373-55241962000100002

Resumo

A fishery research programme requires an accurate picture of the length distribution of the fish caught and landed. Each boat is considered a sampling unit. It is therefore necessary to know how to draw a sample from the fish landed in such a way as to give an accurate assessment of the length distribution of the boat's landing. This is the purpose of the present paper, which refers to the sampling of "pescada-foguete" (Macrodon ancylodon) caught by parejas of "Sociedade de Pesca Taiyo Ltda.", and landed on the fish market of Santos, State of São Paulo, Brazil. By courtesy of that firm two voyages were made during which observations on the length distribution of the fish caught were carried out. Independently, samples were taken of the fish landed from these voyages, after landing. In this way an evaluation of the sampling techniques used on the fish market was obtained. Due to the selection of the nets used, there is no rejection of "pescada-foguete" at sea and therefore the fish measured represent a boat's catch. Methods: 1 - Samples from the hauls were taken on board, at random, during two voyages, to find out if there is a significant difference in the length distributions between hauls on the same voyage. 2 - At the fish market three boxes of each size category: large - G, medium - M, and small - P, were measured: one (1) series at the beginning, the second (2) at the middle, and the third (3) at the end of the landing operation. The purpose was to see if there is a significant difference between the length distributions of each size category or if the same length distribution occurred in a given category during the entire process of landing. 3 - The grand mean, calculated according to the procedure and example given in Table VIII, for the sample quoted in item 2 above, was compared with that taken from the same vessel at sea. This shows whether the market sampling for length distribution differs significantly from that estimated from non-selected samples on the boat. The results were as follows: 1 - In the first voyage seven samples (Table I) were obtained from seven different hauls, and the means of the samples are plotted in Figure 2; the means of the second voyage (fifteen samples) drawn at random on the following trip, are plotted in Figure 3. The t test was applied for the comparison between the means and the results are shown in Tables II and III. The first three means show (Figure 2) a significant difference because at the beginning of the fishing operation several hauls were made at random to locate the shoals (Figure 1). The other four show no significant difference. The means of the second trip do not show, in general, a significant difference, probably due to the fact that the fishing was done at the same distance from the coast during the whole trip. For estimation of the variability of all samples together, analysis of variance was used (Table IV). The value of F is high in the first voyage, but not very high in the second. 2 - The difference between the mean length of the same size category taken at the beginning (1), at the middle (2), and at the end (3) of the landing was not highly significant (Tables V and VI). 3 - Table IX shows the comparison between the grand mean (see how it was computed in Table VIII) and the mean resulting from the gathering of samples obtained aboard the ship, in the first and in the second voyage. Those concerning the first trip do not show a significant difference; the same is true of two [(1) and (2)] from the second trip. The other (3) shows a significant difference because at the end of the landing some fish of bigger size than those normally considered as small were placed in the box of small fish measured. A percentage length frequency distribution of the raised market data and the boat samples (Figs. 4 and 5) shows practically no difference. Also, there is no significant difference between the samples obtained on board and those obtained on the fish market (grand mean for the two stratified samples of the first voyage, and for the three samples of the second one), as shown in Table X, i.e., the mean of two stratified samples taken at the fish market at the return from the first trip, when compared with the one obtained aboard, showed no significant difference. The same is evident with the three stratified samples obtained at the return from the second trip. The size distribution within a category is not the same for every landing (Tables VII and VIII). Therefore, from every landing a stratified sample must be drawn. Table V shows that two general means present no significant difference, but this is not so when compared with the third one. This suggests that more than one box is necessary to represent the catch of a pareja belonging to "Sociedade de Pesca Taiyo Ltda.". Table XI I shows, however, that the measurements of half a box of 25 kg, and the counting of the other half (to raise the total number of fish landed at each size category) are sufficient enough to represent the box. The most important conclusion is that a better representative sampling can be achieved by reducing the number of fish measured from each box, and increasing the number of boxes measured for fish. Thus the variance between boxes will be reduced and it will fall nearer to the variance within each box.

Instituto Oceanográfico da Universidade de São Paulo. Membro do Grupo de Pesquisas sôbre a Pesca Marítima (G.P.P.M.)

SUMMARY

A fishery research programme requires an accurate picture of the length distribution of the fish caught and landed.

Each boat is considered a sampling unit. It is therefore necessary to know how to draw a sample from the fish landed in such a way as to give an accurate assessment of the length distribution of the boat's landing. This is the purpose of the present paper, which refers to the sampling of "pescada-foguete" (Macrodon ancylodon) caught by parejas of "Sociedade de Pesca Taiyo Ltda.", and landed on the fish market of Santos, State of São Paulo, Brazil.

By courtesy of that firm two voyages were made during which observations on the length distribution of the fish caught were carried out. Independently, samples were taken of the fish landed from these voyages, after landing. In this way an evaluation of the sampling techniques used on the fish market was obtained. Due to the selection of the nets used, there is no rejection of "pescada-foguete" at sea and therefore the fish measured represent a boat's catch.

Methods:

1 - Samples from the hauls were taken on board, at random, during two voyages, to find out if there is a significant difference in the length distributions between hauls on the same voyage.

2 - At the fish market three boxes of each size category: large - G, medium - M, and small - P, were measured: one (1) series at the beginning, the second (2) at the middle, and the third (3) at the end of the landing operation. The purpose was to see if there is a significant difference between the length distributions of each size category or if the same length distribution occurred in a given category during the entire process of landing.

3 - The grand mean, calculated according to the procedure and example given in Table VIII, for the sample quoted in item 2 above, was compared with that taken from the same vessel at sea. This shows whether the market sampling for length distribution differs significantly from that estimated from non-selected samples on the boat.

The results were as follows:

1 - In the first voyage seven samples (Table I) were obtained from seven different hauls, and the means of the samples are plotted in Figure 2; the means of the second voyage (fifteen samples) drawn at random on the following trip, are plotted in Figure 3. The t test was applied for the comparison between the means and the results are shown in Tables II and III. The first three means show (Figure 2) a significant difference because at the beginning of the fishing operation several hauls were made at random to locate the shoals (Figure 1). The other four show no significant difference. The means of the second trip do not show, in general, a significant difference, probably due to the fact that the fishing was done at the same distance from the coast during the whole trip.

For estimation of the variability of all samples together, analysis of variance was used (Table IV). The value of F is high in the first voyage, but not very high in the second.

2 - The difference between the mean length of the same size category taken at the beginning (1), at the middle (2), and at the end (3) of the landing was not highly significant (Tables V and VI).

3 - Table IX shows the comparison between the grand mean (see how it was computed in Table VIII) and the mean resulting from the gathering of samples obtained aboard the ship, in the first and in the second voyage. Those concerning the first trip do not show a significant difference; the same is true of two [(1) and (2)] from the second trip. The other (3) shows a significant difference because at the end of the landing some fish of bigger size than those normally considered as small were placed in the box of small fish measured. A percentage length frequency distribution of the raised market data and the boat samples (Figs. 4 and 5) shows practically no difference. Also, there is no significant difference between the samples obtained on board and those obtained on the fish market (grand mean for the two stratified samples of the first voyage, and for the three samples of the second one), as shown in Table X, i.e., the mean of two stratified samples taken at the fish market at the return from the first trip, when compared with the one obtained aboard, showed no significant difference. The same is evident with the three stratified samples obtained at the return from the second trip.

The size distribution within a category is not the same for every landing (Tables VII and VIII). Therefore, from every landing a stratified sample must be drawn. Table V shows that two general means present no significant difference, but this is not so when compared with the third one. This suggests that more than one box is necessary to represent the catch of a pareja belonging to "Sociedade de Pesca Taiyo Ltda.". Table XI I shows, however, that the measurements of half a box of 25 kg, and the counting of the other half (to raise the total number of fish landed at each size category) are sufficient enough to represent the box. The most important conclusion is that a better representative sampling can be achieved by reducing the number of fish measured from each box, and increasing the number of boxes measured for fish. Thus the variance between boxes will be reduced and it will fall nearer to the variance within each box.

Texto completo disponível apenas em PDF.

Full text available only in PDF format.

VI - AGRADECIMENTOS

O autor agradece a orientação dada por Mr. Ian Dennis Richardson, técnico da FAO, e as sugestões apresentadas pelo Eng. Antônio Garcia Occhipinti, meteorologista do Instituto Oceanográfico da USP. Os agradecimentos são extensivos à Sociedade de Pesca Taiyo Ltda., proprietária do barco "Akashi Maru 33", no qual viajou o autor, e em especial ao Capitão desse barco, Sr. Shigeo Kobayashi, pela cordial acolhida a bordo e informações prestadas sobre técnicas de pesca. São também credores da gratidão do autor todos os membros do Grupo de Pesquisas sobre a Pesca Marítima que colaboraram na execução deste trabalho.

VII - BIBLIOGRAFIA

(Recebido em 8/5/61)

Publ. nº 159 do Inst. Oceano da USP.

Con tr. nº 10 do G.P.P.M.

CROXTON, F. E. & COWDEN, D. J. 1952. Estatística geral e aplicada. Rio de Janeiro, Inst. Bras. Geog. Est. xxiii+1096 p., 257 graf.
POPE, J. A. 1956. An outline of sampling techniques. Rapp. Proc. - Verb. Réun., vol. 140, Part I, p. 11-20.