POWER and McCLEAVE: SIMULATION OF THE NORTH ATLANTIC DRIFT OF ANGUILLA 



ing concentration. No mortality of leptocephali was 

 incorporated into the model, and the total number of 

 leptocephali was conserved throughout the simula- 

 tion except for the portion transported across the 

 open boundaries discussed previously. A unitless 

 number is given when referring to a concentration, so 

 that reference to a concentration of 10~ 3 refers to a 

 concentration of leptocephali that is three orders of 

 magnitude below the starting concentration. The 

 concentration contours presented in Figures 3-9 

 were determined by linearly interpolating between 

 the concentrations at the centers of the cells. Each 

 contour represents an order of magnitude change in 

 concentration relative to neighboring contours. Only 

 leptocephalus concentrations >10" 7 (proportion of 

 the starting concentration) are shown. In Figures 3-9 

 the leptocephalus starting location is marked by a 

 star. 



The choice of contour intervals as orders of 

 magnitude was arbitrary; in some cases, the display 

 masked the spatial structure of the distributions. 

 This can occur where the order of magnitude con- 

 tours are widely spaced, and a discontinuity in the 

 concentrations is between the contours. For this 

 reason an agglomerative cluster analysis using a spa- 

 tial autocorrelation coefficient, Moran's I (Cliff and 

 Ord 1973), as the metric was carried out with cell con- 

 centrations as the variable. The weighting coefficient 

 was the reciprocal of the distance between cells, and 

 only immediately adjacent cells were linked (rook's 

 moves). Examination of the equation for Moran's I in- 

 dicates that when choosing among several coefficient 

 values, the minimum I represents the most spatially 

 uniform distribution. The clustering proceeded 

 iteratively by examining all possible pairwise link- 

 ages of clusters of cells, and forming a new cluster 

 from the pair that yielded the minimum value of I for 

 the new cluster. Thus at any stage the clusters parti- 

 tion the distribution into "patches," i.e., regions in 

 which the cells are most spatially uniform in concen- 

 tration. The cell concentrations were logarithmically 

 transformed for the analysis to minimize the effects 

 of outliers on I. The results of these analyses are not 

 explicitly presented, but are referred to when 

 necessary to facilitate the interpretation of the con- 

 tour plots. 



RESULTS 



Distribution of Leptocephali After 

 30 Days of Drift 



The proximity of the starting point to the Gulf 

 Stream, the Antilles Current northeast of the 



Bahamas, and the Bahamas themselves all influenced 

 the distributional patterns of American eel lep- 

 tocephali that developed 1 mo after the 1 March 

 starting date (Fig. 3). Capture of leptocephali by the 

 Gulf Stream was already evident, and larvae started 

 east and northeast of the Bahamas showed north- 

 westerly drift on the Antilles Current. 



Leptocephali begun at F mostly moved away from 

 that location during the first month (Fig. 3F). The lar- 

 vae were somewhat dispersed even at this early date, 

 as cluster analysis indicated a large patch east of the 

 Bahamas between lat. 21° and26°N. Concentrations 

 declined sharply to the east and south, as they did in 

 all runs, indicating little transport in those directions. 

 To the west, leptocephali were split by the Bahamas, 

 approaching the Gulf Stream by both the route north 

 of the islands and through the channel between the 

 Bahamas and Cuba. Some passed completely 

 through this channel to be caught in the Florida 

 Current and carried northward, so that the 10~ 7 con- 

 tour extended to lat. 32°N. 



There was gradual northwesterly dispersal of lep- 

 tocephali started at more northerly (Fig. 3B, D) and 

 northwesterly (Fig. 3A, C) locations than those at F. 

 The larvae in these runs are more concentrated, with 

 most remaining near the starting points. In runs A-D 

 larvae are impinging upon the Bahamas, and in C the 

 concentrations offshore of the Bahamas are par- 

 ticularly high. This is due to the clearly evident An- 

 tilles Current transport in run C, and this current also 

 facilitated the entry of run C larvae into the Gulf 

 Stream. Runs A and E also show Gulf Stream 

 transport, but this is more by virtue of their starting 

 point's proximity to the Gulf Stream. Gulf Stream 

 transport is pronounced in E, with the 10" 7 contour 

 reaching north to lat. 39°N and east to long. 65° W. 



Distribution of Leptocephali After 

 60 Days of Drift 



By 30 April, 2 mo after the 1 March start, lep- 

 tocephali had spread and most moved northwest of 

 their starting locations (Fig. 4). There were now 

 broader areas of more moderate concentration, 

 typified by the area enclosed by the 10" 2 contour east 

 of the Bahamas. Gulf Stream transport was now evi- 

 dent in all runs A through F. 



Run F is notable for the substantial distance 

 traversed by the larvae, considering the position of 

 point F (Fig. 4F). This was primarily caused by con- 

 tinued transport between the Bahamas and Cuba, 

 entry into the Florida Current, and then rapid Gulf 

 Stream transport. Antilles Current transport also 

 contributed. Concentrations in run F fell into three 



487 



