FISHERY BULLETIN: VOL. 83, NO. 3 



to diffuse the Jight. Four light sources were equally 

 spaced around the periphery of the tank; the top of 

 each being just below the tank rim. Each source con- 

 sisted of a 30 W tungsten microscope lamp with 

 reflector enclosed in a tube. On top of the tube were 

 two color filters; a green acrylic plastic filter (#2414, 

 Rohm and Haas), and a sealed petri dish containing a 

 5% CUSO4 solution (by weight). To diffuse the light a 

 translucent white acrylic filter (#W-2447, Rohm and 

 Haas) was placed on top of the color filters and light 

 transmitted by the white filter entered a white opal 

 glass globe (13 cm diameter) that formed the top of 

 the source. The lamps were operated for 5 h (10% of 

 lamp life) before they were used to produce test 

 levels of irradiance. These sources were used to pro- 

 duce four test levels of irradiance during a 12-h 

 night. Additionally, four tungsten 100 W household 

 lamps (unfiltered) with reflectors were used for the 

 daytime level of irradiance. These lamps were placed 

 at regular interv-als around the perimeter of the tank 

 near the ceiling. Light from these lamps reflected off 

 the ceiling providing a uniformly diffuse illumination. 

 The spectrum produced by the four sources (Table 1) 

 resembles the greenish spectrum tj^jical of anchovy 

 habitat, but the spectrum used in the day was not dif- 

 ferent from a standard curve for a tungsten lamp 

 and consequently contained an unnaturally high pro- 

 portion of longer wavelength energy. Our term for 

 the condition when all lamps were off was darkness; 

 under these conditions light was not detectable by a 

 dark adapted human observer and the irradiance was 

 below the sensitivity of a 931 A photomultiplier 

 which can detect about 5 x lO-*' mc (meter candle). 

 To record the effect of light on the schools the fish 

 were photographed from above the tank using a 35 



mm automatic camera and flash attachment. The 

 camera was controlled by a timer, and photographs 

 were taken at 30-min intervals for 5 h during the 

 12-h day, at night in darkness, and at night at the 

 test levels produced by the four sources. Night 

 photographs were taken during a 5-h period com- 

 mencing 2 h after the end of the 12-h day. Ten 

 photographs were usually analyzed at each light level 

 for each group, but in several tests, 1 or 2 photo- 

 graphs were not analyzed because not all 50 fish 

 could be seen. 



Two indices of schooling were calculated for each 

 photograph: an index of dispersion (Pielou 1969), and 

 the mean distance to the nearest neighbor (Hunter 

 1966). The dispersion index was calculated by super- 

 imposing a grid containing 326 quadrats over the 

 projected image of the tank and counting the number 

 of fish occurring in each quadrat. The variance mean 

 ratio (s'^/x) for the number of fish per quadrat was the 

 index of dispersion. The index was calculated for 

 each photograph, and an average index was com- 

 puted for each light treatment (n = 8-10 photo- 

 graphs). A dispersion index of 1 indicates a random 

 distribution, whereas higher values indicate aggrega- 

 tion (Pielou 1969) and imply the existence of school- 

 ing. Values < 1 imply a uniform distribution over the 

 grid. The mean distance to the nearest neighbor was 

 computed for a random subsample of 10 fish in a 

 photograph. All 50 fish in a photograph were 

 numbered and the subsample of 10 was selected by 

 drawing the fish numbers from a table of random 

 numbers. For each of the 10 fish in the subsample 

 the distance in centimeters to its nearest neighbor 

 was measured (distance between heads), a mean 

 distance calculated for each photograph, and means 



Table 1.— Spectral functions used to estimate the depth of occurrence of the 

 visual threshold for schooling under the various water types and incident irra- 

 diances including spectral irradiance in the laboratory apparatus, moonlight at 

 3 m below water surface, starlight at the earth's surface (Munz and McFarland 

 1977), and the relative sensitivity of the dark adapted anchovy eye (Engraulis 

 encrasicholus, Protasov 1964). 



'Wavelength Interval = 588-600 nm. 



236 



