SQUIRE: ANCHOVY SCHOOL SHAPES 



School length to width ratios were determined 

 for both day and night schools. 



The school area, as expressed in terms of 

 percentage of the circle drawn about it, was de- 

 termined by tracing the school profile upon paper, 

 cutting the school out of the circle and weighing 

 both the school profile and the nonschool portion of 

 the circle on a sensitive laboratory balance. School 

 areas for the six points of school width (diameter) 

 measurement about the 180° arc were computed 

 statistically. 



RESULTS 



Night Observations 



Figure 1 illustrates the profiles of schools re- 

 corded by the airborne low-light-level television 

 system. On each figure are given the area percent- 

 age of a tangent circle that the fish school occupies 

 and the length to width ratio. 



The average percent coverage of a night an- 

 chovy school, in relation to the circle tangent to 

 the school, is 29. 27^ The average ratio of school 

 length to width is 2.53:1. 



Day Observations 



Figure 2 illustrates profiles of schools observed 

 during the day. On each figure are given the per- 

 centage of a circle tangent to the school that is 

 occupied by the fish school, the length to width 

 ratio, and the 30° arc points that were used to 

 determine the school's estimated diameter and 

 area, simulated as randomly viewed by a side- 

 looking sonar. The ratio of the actual school area 

 to the calculated area of the school's average, high 

 and low estimate, as viewed every 30° of a 180° arc 

 based on simulated sonar measurements of width, 

 is given in Table 1. 



The average percent coverage of a day anchovy 

 school to the tangent circle about the school is 

 \2.V7c. The average length to width ratio for all 

 day schools is 2.09:1. The ratio of estimate of the 

 area of all day schools, as calculated from mea- 

 surements from 30° arc points about the school, to 

 the actual area of the school is 1.72:1. The ratios, 

 length to width, and estimate of school area to 

 actual school area were compared and Figure 3 

 graphs the relationship. The graph displays the 

 variables plotted on log-log paper showing two 

 main points: One, that the variance is changing 

 proportionally to the mean. This is expected as 

 there should be more variation as the school 



Table l. — Ratio of the actual anchovy school area to the aver- 

 age area based on six points of observation as viewed every 30° of 

 a 180° area, and the high and low ratio. 



length to width increases. Two, the plotted regres- 

 sion line indicates that more bias (higher esti- 

 mated actual school ratio) is introduced as the 

 school length to width ratio increases. The line is 

 significant at the 95'7f confidence interval as 

 proven by the ^test ( 1.98 <2.298). The confidence 

 limits are from 0.0545 to 0.734. 



SUMMARY AND COMMENTS 



The data on day/night school length to width 

 ratios support what is commonly known about the 

 schooling shapes of the northern anchovy. They 

 are more common in the near-surface area at 

 night, generally in large elongate thin surface 

 schools. These elongate schools tend to group to- 

 gether in the early morning hours and descend to 

 depth to form more compact schools during the day 

 (Mais 1974). Studies by Squire ( 1972) of aerial fish 

 spotter data show anchovy schools to be more fre- 

 quently observed, and observed in larger quan- 

 tities at night, when compared with day observa- 

 tions. 



The schools percent area coverage of the tangent 

 circle at night is 12.9*^ less than its coverage dur- 

 ing the day and the length to width ratio is greater 

 by 0.44. In addition, analysis of school length to 

 width ratios compared to the ratio of estimated 

 school size to actual size (Figure 3) shows that as 

 the length to width ratio increases a greater error 

 in school area estimate will occur. Schools with a 

 length to width ratio of 2:1 have an estimated to 

 actual error of about 1.5:1 while more elongate 

 schools of a ratio of 3:1 have an estimated error of 

 about 1.75:1. 



445 



