FISHERY BULLETIN: VOL. 81, NO. 3 



30 May). Two months later the 10" 2 enclosed about 

 the same area as before, but it had moved a few 

 degrees to the northwest of the starting location (Fig. 

 9B; 29 July). The lower concentrations continued to 

 spread, and dispersal occurred in most compass 

 directions. By day 165, the lower concentrations had 

 expanded still further, although concentrations 

 >10 -2 maintained approximately the same position 

 (Fig. 9C; 27 September). This same pattern of dis- 

 persal continued to day 215 of drift, at which point 

 the concentrations <10" 5 at the northern limits of 

 the distribution showed signs of being captured by 

 the Gulf Stream (Fig. 9D; 26 November). The simula- 

 tion was halted at day 215. 



DISCUSSION 



The simulations revealed several previously unsus- 

 pected features of the leptocephalus drift migration, 

 such as the patch formation offshore of Florida and 

 the Gulf Stream. This is in spite of the simplifications 

 and assumptions that were made in a model encom- 

 passing such a large geographic area. The bound- 

 aries, currents, and eddy diffusivities all provided 

 only an approximation to the physical system. None- 

 theless, interpreting the simulation output using the 

 available information on leptocephalus distribution 

 and the eel's life history indicates that the model has 

 realistically reproduced the large-scale features of 

 the drift migration. 



The simulated drift of American eel leptocephali 

 can be divided into four phases, the first being initial 

 northwest transport following spawning. This trans- 

 port was largely on the Antilles Current, which flows 

 northwesterly on a course parallel to the north- 

 eastern border of the Bahama Islands chain (Fig. 2). 

 The extent of the initial larval transport depended on 

 the starting location's position with respect to this 

 current. Larvae started farther northeast in the 

 Sargasso Sea (point B and other simulations not 

 shown here) showed less unidirectional movement 

 than those started in or nearer the Antilles Current. 

 This current clearly appeared in ships' drift data for 

 May (Fig. 2) and other late winter and spring months, 

 so its effects in the simulations were no surprise. How- 

 ever, there are questions concerning the existence 

 of the Antilles Current (Ingham 1975; Gunn and 

 Watts 1982). Gunn and Watts (1982) showed that the 

 Antilles Current was present in January- February 

 1973, but that the region was dominated by eddies in 

 July- August 1972. They speculated that the Antilles 

 Current may only exist seasonally. Its presence in 

 January-February 1973 must have dominated the 

 drift of newly hatched leptocephali from early spawn- 



ing in that year. If the Antilles Current is seasonal, the 

 question important to eel biology is how long does it 

 persist through the spring and summer? The Antilles 

 Current was important in the simulations from the 

 time of spawning up to May-June. After that its in- 

 fluence on the distribution of 0-group leptocephali 

 diminished, and by July-August, when Gunn and 

 Watts (1982) did not find the current, most of the lar- 

 vae were north of lat. 25°N. A study like that of Gunn 

 and Watts is clearly needed for the months between 

 February and July, and particularly for April and 

 May. 



There was little eastward transport from the start- 

 ing locations in all American eel simulations, includ- 

 ing those in which larvae were started along long. 

 66°W (not pictured in this paper). Schoth and Tesch 

 (1981) collected American eel leptocephali east of 

 long. 69°W (the longitude of the easternmost starting 

 points in the simulations presented here), although 

 the numbers of larvae they caught declined rapidly 

 east of long. 65°W. Transport to the south was also 

 minimal in the simulations, and few leptocephali en- 

 tered the Caribbean. The simulations do not ade- 

 quately explain the presence of leptocephali that 

 have been collected in the Caribbean and Gulf of 

 Mexico, and in particular the presence of young lep- 

 tocephali near the Yucatan peninsula (Kleckner and 

 McCleave 1980) was not reproduced. American eel 

 spawning in the Caribbean remains a viable explana- 

 tion for these collections. 



The simulated drift of leptocephali south and west 

 of the Bahamas must be interpreted cautiously, 

 because this region's complicated bathymetry and 

 currents were not well represented in the model. 

 However, there were some striking correspondences 

 between the simulations and the actual collection 

 data for this region. Smith(1968) reported 10 Mayas 

 the earliest collection date for 0-group leptocephali 

 in the Straits of Florida between the Bahamas and 

 Florida; Figure 3F shows larval concentrations of 

 10 -6 (proportion of the starting concentration) had 

 arrived in the Straits of Florida around 30 March, and 

 by 30 April the concentration had increased to 10" 4 

 (Fig. 4F). Kleckner and McCleave (1980) reported 

 the collection of 0-group leptocephali in the Bahama 

 Island chain in April to June, and the first collections 

 of larvae in the Straits of Florida in May. Smith 

 (1968) gave 28 August as the latest collection date of 

 0-group larvae in the Straits of Florida, and by that 

 date most of the leptocephali in the simulations had 

 departed the area as well. 



The formation and maintenance of a patch of lep- 

 tocephali north of the Bahamas and east of Florida 

 and the Florida Current-Gulf Stream system were the 



496 



