FISHERY BULLETIN: VOL. 75. NO. 2 



the twilight transitional periods. This would be 

 the predicted relationship since fish respond to 

 specific intensities and spectral composition of 

 light (Munz and McFarland 1973). The intensity 

 and spectral composition of incident light at 

 specific times relative to sunrise or sunset are 

 identical each day, although they vary with time 

 and season and with latitude. The amount of cloud 

 cover and/or high mountainous terrain nearby, as 

 in Kaneohe Bay and Kona, Hawaii (Hobson 1972) 

 or Baja California (Hobson 1968), may shift the 

 activity patterns to later in the morning, or earlier 

 in the evening (i.e., shift the time relative to 

 sunrise and/or sunset at which specific light levels 

 occur). However, the basic relationships between 

 behavior and twilight periods appear to hold. 



Light meter readings recorded during the 

 formation and break up of Hawaiian silverside 

 schools are compared with those recorded for two 

 other species of siversides in Table 1. The 

 readings for all three species are not significantly 

 different. Such light levels occur naturally when 

 the sun is between -5° and —9° below the horizon 

 during the periods of evening or morning twilight 

 (Brown 1952). These data and the field observa- 

 tions reported here are also comparable to the 

 light levels and the sun angles calculated from the 

 data presented by Pavlov (1962) for another 

 silverside, Atherina mochon pontica. Pavlov found 

 that peak predator success occurred at light levels 

 of approximately 0.01-108 foot candles corre- 

 sponding to sun angles of - 9° to + 1 ° to the horizon 

 (Brown 1952) (i.e., centered during the period of 

 civil twilight). 



These comparisons indicate that related species 

 of silversides, which live in widely separate parts 

 of the world, have similar visual thresholds and, 

 perhaps, sensitivity. Munz and McFarland (1973) 

 provided a synopsis of research, which has shown 

 that many related species demonstrate a consid- 

 erable diversity in their visual sensitivity. How- 

 ever, species, whether related or not, which occur 

 in similar environments, appear to have similar 

 thresholds and sensitivity. These relationships 

 indicate that the above silverside species from 

 various locations in the world may have very 

 similar behavioral patterns and/or live in very 

 similar physical and biological environments. 



When light levels decrease in the evening, 

 visual thresholds may be reached, making coordi- 

 nated schooling movements impossible, or at least 

 more difficult for the silversides. These thresholds 



may be reached at the time when cone vision shifts 

 to rod vision (the Purkinje shift), neither cone nor 

 rod vision being fully efficient (Munz and McFar- 

 land 1973). As school formation breaks down or 

 increases, the silversides appear to be the most 

 vulnerable to predatory attack. This vulnerability 

 may be due to reduced visual sensitivity, leading 

 to an inability to see their predators below them 

 against a dark bottom or deep water (Hobson 

 1966, 1968) and react in time to avoid and escape 

 from them (Dill 1972, 1974a, b). In addition, such 

 prey may be unable to simultaneously interact 

 with conspecifics, and look out for predators at a 

 distance at low light levels. 



Predators are presumably able to see their prey 

 at a horizontal angle or silhouetted against the 

 twilight sky for a short period of time before their 

 lower visual threshold is reached in the evening 

 (Hobson 1966, 1968). Munz and McFarland ( 1973) 

 indicated that increased visual sensitivity in 

 predators, which provides sufficient resolution for 

 the detection of prey in motion during twilight, 

 may be a result of having relatively larger, but 

 fewer, cones in their retinas compared with those 

 found in diurnal fishes. This factor is critical since 

 predators must align themselves and be able to 

 predict where their prey will be during the mouth 

 opening phase of their strike (Nyberg 1971). 



Weighing against the hypothesis that the 

 schools of silversides break up and reform as a 

 result of changes in visual sensitivity, are a 

 number of observations made of captives held in 

 the field enclosures in the absence of predators. 

 When held for weeks at a time, these silversides 

 did not completely lose their cohesion and 

 polarity, indicating that there may be a strong 

 genetic component to their schooling behavior. 

 This genetic component may result in the silver- 

 sides remaining within a short distance of one 

 another at all times. The silversides appear to be 

 adapted to feeding at night as well as in the day 

 (McMahon 1975). If they can feed at night, the 

 silverside are probably able to detect the presence 

 of conspecifics, either using visual and/or lateral 

 line cues. The ability to detect conspecifics would 

 be particularly beneficial as individuals would not 

 become so widely scattered during the night that 

 polarized schools could not easily reform during 

 morning twilight. In addition, the observation 

 that captive silversides held in large enclosures in 

 the field in the absence of predators did not all 

 spread out to look continuously for food indicates 



424 



