FISHERY BULLETIN: VOL. 76. NO. 3 



methods of selecting transects, or allocating sam- 

 pling effort. The three types of simulated survey, 

 with a range of sample sizes (numbers of tran- 

 sects), were run on 15 model anchovy populations. 



METHODS 



Anchovy populations were modeled as arrays 

 with each element representing 1 n.mi.-. The 

 ari-ay dimensions were 180 x 75, approximately 

 the dimensions, in miles, of the Los Angeles Bight. 

 Since a school is the population unit detected in an 

 acoustic or aerial survey, the units of the model 

 populations were schools. One hundred fifty 

 thousand schools were distributed in the array 

 resulting in a mean density of 11.1 schools mi^. 

 Four acoustic surveys by the California Depart- 

 ment of Fish and Game- in 1975 and 1976 yielded 

 estimates ranging from 88,887 to 319,878 an- 

 chovy schools off southern California in the area 

 of the bight. Mais (1974) gave a range of 21,920- 

 343,070 ix = 150,996) schools off southern 

 California and northern Baja California, most of 

 which were within the bight. 



The schools were placed in circular school 

 groups located at random. Schools were distrib- 

 uted uniformly within a school group. School 

 group radii and densities were chosen randomly 

 and independently from log-normal approxima- 

 tions of observed frequency distributions based on 

 52 school groups from six California Department 

 of Fish and Game Sea Survey acoustic surveys 

 (MacCall et al.-M (Figure 1). There was no sig- 

 nificant correlation between the density of schools 

 within a school group and the size of the school 

 group in these observations. Where school groups 

 overlapped, the densities were simply added to- 

 gether, although this effectively increased both 

 the mean radius and density. In one model popula- 

 tion illustrated in Figure 2, 16 school groups con- 

 taining 150,303 schools covered about 149^ of the 

 survey area. Fifteen model populations were used, 

 each with the same total number of schools, but 

 different locations, sizes, and densities of school 

 groups. 



^S.J.Crooke. 1975. Cruise reports 75- A- 1 and 75- A-6. K. 

 F. Mais. 1976. Cruise reports 76-A-3 and 76-A-9. State of 

 California - The Resources Agency, Department of Fish and 

 Game, Marine Resources Region, Long Beach, CA 90802. 



■■'MacCall, A., P. E. Smith, G. Stauffer, J. Squire, J. Zweifel, 

 and S. Crooke. Report of CalCOFI anchovy workshop working 

 group on methods of estimating anchovy abundance. Unpubl. 

 manuscr. Southwest Fisheries Center, National Marine 

 Fisheries Service, NOAA, La Jolla, CA 92038. 



2.2 2.6 3.0 3.4 38 4.2 4.6 5.0 54 

 Lege DENSITY (schools /miles^) 



0.40 



Log e RADIUS (miles) 



Figure l. — Comparison of distributions of northern anchovy 

 school group density and size observed in the California Current 

 during California Department Fish and Game surveys (solid 

 lines) to log-normal approximations used in simulations (dashed 



lines). 



A simulated survey consisted of a series of 

 transects across the survey area. There were 180 

 possible transects, each 1 mi wide. Acoustic sur- 

 veys currently run by the Southwest Fisheries 

 Center, National Marine Fisheries Service, 

 NOAA, used a transect width of 0.14 mi (250 m). 

 Aerial transect widths were typically 0.2 to 0.5 mi. 

 A larger transect width was used in the simula- 

 tions to hold the model population array down to a 

 reasonable size. We assumed that the general re- 

 sults of the simulations would not change by using 

 a smaller transect width. Since all schools were 

 coimted within a transect, the only source of error 

 in the survey estimate was the large variance in 

 the number of schools per transect. For instance, 

 in the model population in Figure 2, the mean 

 number of schools per transect was 835.0, while 

 SD was 920.4 (variance = 8.47 xlO^). 



Systematic surveys were simulated by counting 

 the schools within a series of transects separated 

 by a constant transect interval. A population es- 

 timate was calculated simply by dividing the 

 survey count by the fraction of the survey area 

 covered by the transects. Transect intervals of 2, 3, 



680 



