FISHERY BULLETIN: VOL. 75, NO. 3 



behavior which still resulted in gamete release. 

 For example, as the female was moving to the 

 surface, rather than orienting the ventral area of 

 her body toward the male, she bent her body into 

 the U-shape with the result that her dorsal side 

 faced the ventral side of the male. Gamete release 

 still occurred as the fish flexed their bodies into 

 U-shapes. The origin of this variation was usually 

 due to the fact that while moving upward the 

 female was swimming too rapidly to assume the 

 proper alignment for the ventral-to-ventral re- 

 lease with the male. 



A critical factor for maximizing fertilization 

 was the breaking of the water surface at the time 

 of release. As the fish moved upwards, churned the 

 water, and swam downwards again, currents were 

 created which mixed the "cloud" of gametes to- 

 gether. From visual observations and motion pic- 

 ture analysis, this occurred whether there was 

 ventral-to-ventral or ventral-to-dorsal alignment 

 of the pair. We would assume, however, that the 

 most efficient method for fertilization involved the 

 ventral-to-ventral alignment. 



While runs were always performed prior to 

 spawning, on some days there were as few as 2 

 runs prior to a spawning, while on other days there 

 were as many as 11. Similarly, the duration of a 

 series of runs varied from 30 to 180 s. 



Runs were not always performed in succession. 

 Particularly in Study 1, many times after complet- 

 ing one run, the pair began circling around each 

 other in midwater. In some cases they followed 

 each other, head to tail, along the perimeter of an 

 imaginary circle. In other cases, as the male swam 

 around the female, she either remained sculling in 

 a fixed position or pivoted about her vertical axis, 

 obviously orienting to the moving male. The total 

 number of separate circling bouts during a run 

 sequence ranged from 2 to 10 with a duration of 

 each ranging from 2 to 40 s. 



Occasionally at the end of a run, the pair began 

 to swim upwards, as if to spawn. Typically, at the 

 onset of this, the female began the transition from 

 pectoral swimming to caudal thrusts. Moving 

 rapidly upwards with the male alongside, the 

 female broke away from him short of the surface 

 and swam downwards to the sand without releas- 

 ing gametes. This behavior sometimes did not 

 occur at all while in other cases it occurred as 

 many as six times prior to a spawning. 



The continuity or fluidity of the run sequences 

 appeared to be a critical factor serving to syn- 

 chronize the fish for final release of gametes. Lack 



594 



of mutual stimulatory behaviors or even slightly 

 inappropriate behavior by one of the mates during 

 a run, in general, were sufficient causes for a tem- 

 porary breakoff of the entire sequence. During a 

 breakoff the female's shading often regressed 

 somewhat and she came to rest on the sand or even 

 returned to the shelter for a few seconds. 



One of the specific causes for these breakoffs was 

 due to the fact that the dominant male, instead of 

 maintaining his attention toward the female, 

 chased or displaced the subordinate male which 

 had either ( actively) moved too close to the pair or 

 (passively) happened to be in areas where the pre- 

 spawning behavior was being carried out. 



Other reasons for the breakoffs were inappro- 

 priate stimuli initiated usually by the male during 

 the run sequence. In Study 1, if the male contacted 

 the female during a run along the sand or as she 

 ascended to spawn rather than at the apex of the 

 pathway, the female often turned away from the 

 male. Conversely, premature contact behavior by 

 the male in Study 2 was an appropriate stimulus 

 to his mate and in fact was frequently exhib- 

 ited during the run sequence as well as during 

 spawning. 



Other cases in which the female initiated a 

 breakoff from a run occurred if the male assumed 

 an atypical position relative to hers. In Study 1, 

 the female usually swam between the wall and the 

 male and slightly ahead of him. Occasionally if the 

 male assumed the position closest to the wall dur- 

 ing a run (i.e., the female was now closer to the 

 center of the tank) or if the male swam ahead of 

 her, the female broke away. Since the male some- 

 times "corrected" his position relative to hers and 

 hence the female did not break away, it appeared 

 that each animal had become conditioned to a 

 rather stereotyped set of behavioral patterns and 

 positions which facilitated bringing the spawning 

 to completion. 



Reproductive Behavior of 

 the Subordinate Male 



In both studies each subordinate male had 

 achieved gonadal maturation and was able to 

 complete spawning with the female under a lim- 

 ited set of conditions. In each case, the reproduc- 

 tive behavior occurred only when the subordinate 

 was not behaviorally inhibited by the dominant 

 male. In Study 1, the first spawning by the subor- 

 dinate male and the female occurred later in the 

 spawning season, on the very day (29 May 1975) 



