FISHERY BULLETIN: VOL. 74, NO. 4 



shrimp captured by the iih tow and ^,. ^ be the 

 amount of time tow / spent in the Jth depth zone. 

 Then the proportional catch c,. ^ from the iih tow in 

 thejth depth zone is 



Ci.j = 



ki 



(1) 



For each depth zone ./, summing proportional 

 catches from all tows and dividing by total 

 trawling time in the zone gives the catch rate /• : 



rj = 





(2) 



Ideally, the catch rate is proportional to the 

 population density, so that dividing the catch rate 

 by trawl filtering rate gives an estimated popula- 

 tion density; i.e., 



D, = 



M, •/• V 



(3) 



where D, is the estimated population density in the 

 /th zone, M,. is the effective mouth area of the trawl 

 (because of the design of the trawl, this quantity 

 decreases with increasing towing speed),,/" is the 

 filtering efficiency of the trawl, and r is the towing 

 speed. 



Proportional allotment of the catch by this 

 method assumes that a particular shrimp is equal- 

 ly likely to have been captured at any instant 

 during the tow. This assumption is clearly false for 

 tows that spend only part of their time in the 

 shrimp's actual depth range. However, spurious 

 catch rates outside the actual depth range are 

 minimized by additional tows in these zones that 

 do not enter the actual depth range and do not 

 catch shrimp; these tows increase the denominator 

 of Equation (2) without increasing the numerator. 

 It follows that this method of estimating vertical 

 distributions works best when each depth zone is 

 sampled many times. 



Table 2 shows that during the daytime all depth 

 zones between 400 and 1,075 m were sampled at 

 least five times and that at least five tows spent 

 more than 10 min in all zones between 425 and 950 

 m. Nighttime sampling was less thorough because 

 tows were split into two groups on the basis of 

 moonlight. In both groups all zones in the upper 

 200 m were sampled at least five times, as was the 

 600- to 700-m range (NIGHT) and 650- to 725-m 



range (MOON). NIGHT tows in the 200- to 225-m 

 zone sampled only 2 min; estimated population 

 densities for this zone, while generally plausible- 

 looking, should be regarded cautiously. The 0- to 

 25-m zone for MOON tows were sampled many 

 times for brief periods by open tows that spent 

 nearly all their time at depths of 50-150 m, but was 

 never sampled extensively by any tow. Many 

 species show spuriously high estimated population 

 densities in this zone. There were no NIGHT tows 

 between 1,150 and 1,300 m, and no MOON tows 

 between 375 and 400 m or below 1,175 m. Night- 

 time sampling was generally sparse below 800 m, 

 and the estimated population densities for this 

 region are very crude. 



A second major assumption of this method of 

 presenting vertical distribution data is that the 

 vertical distribution remains constant throughout 

 the sampling period, allowing data from many 

 different cruises to be summed together. The 

 resulting estimated population densities repre- 

 sent an average over the entire sampling period. 

 The actual vertical structure on any given cruise 

 may vary considerably from this average. The 

 separation of nighttime tows into NIGHT and 

 MOON tows is the only systematic attempt to 

 show variations in vertical distribution; other 

 variations are discussed in the species accounts. 



Presentation of Results 



A brief explanation will aid in interpreting the 

 vertical distribution figures that follow (e.g., 

 Figure 1). Catch rates were converted to estimated 

 population densities in numbers per 10"' m' by 

 assuming an average trawling speed of 2 m/s, 

 effective trawl mouth area of 5.1 m- (at 2 m/s), and 

 filtering efficiency of 90%. DAY, NIGHT, and 

 MOON (see above) distributions are shown for the 

 entire population as histograms on the right side 

 of the figure. The number to the right of each 

 histogram is the sample size. In addition, the 

 catches were divided into size classes, and popula- 

 tion densities were estimated by the same method 

 for each size class. Species with a maximum 

 carapace length less than 17.0 mm were divided 

 into 0.5-mm classes, while larger species were 

 divided into 1.0-mm classes. The result was an 

 array of estimated population densities as a 

 function of size and depth. Interpolation produced 

 a series of contours of equal population density. 

 The lowest contour level represents 0.2 shrimp per 

 10"' m"* per mm CL; each successive contour level 



804 



