SOMERTON AND KOBAYASHI: FISH LARVAE CATCHES IN PLANKTON NETS 



160 

 120 



80 



LlJ 



< 



q: 40 

 < 



O 

 q: 



m 





 60 



40 



20 



DAY 

 NIGHT 



UMh. 





 2.0 4.0 6.0 



8.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0 

 LENGTH (mm) 



Figure 3. — The number of larvae by 0.5 mm length intervals captured in Pearl Harbor, HI, 

 in March 1988. The entry experiment (upper panel) used a 0.335 mm standard mesh net 

 during the day and night. The retention experiment (lower panel) used both a 0. 183 mm 

 mesh test net and a 0.335 mm standard mesh net during the day. 



Equation (5) or (6) could be simplified by delet- 

 ing either P,JP,-e or P^JPer, but parameter esti- 

 mation would still require a simultaneous fit of 

 the two equations to the catch ratios. A second 

 example occurs when only an entry e.xperiment 

 or retention experiment is conducted and the 

 appropriate assumption is violated. In this case, 

 the entry and retention probabilities still must 

 be expressed as functions of larval length, thus 

 requiring that nonlinear regression be used to fit 

 either Equation (5) or (6) to the catch ratios. 



The efficacy of the standard net at sampling 

 nehu larvae can be judged in two ways. First, it 

 can be judged by the length range sampled with 

 a P(. = 1.0; that is, the range that requires no 

 correction for extrusion and avoidance. For the 

 standard net, P,. reaches a maximum of 0.86 at 

 4.25 mm and remains above 0.75 only over the 

 interval .3.75-5.50 mm (Fig. 1). In other words, 

 no interval within the larval length range of nehu 

 (2.5-25.0 mm) is sampled completely with the 

 standard net. 



A second way of judging the efficacy of the 

 standard net is by the length range than can be 



corrected, with sufficient precision, for extru- 

 sion and avoidance. The effect of correcting a 

 large sample of nehu length frequencies for ex- 

 trusion and avoidance can be seen in the fre- 

 quency distributions before and after A^„ was 

 divided by the estimated value of Pc (Fig. 4). 

 The precision of this correction can be gauged 

 from the estimates of the variance of A^ (Fig. 5). 

 Note that variance increases gradually with 

 length until 6.75 mm and, thereafter, increases 

 at a greatly accelerated rate. If 6.75 mm is 

 chosen as the upper bound on the length interval 

 within which the estimated numbers are con- 

 sidered sufficiently precise, then only one-third 

 of the larval length range could be corrected to 

 reflect the true length distribution. Thus, judg- 

 ing from either perspective, the standard net is a 

 relatively ineffective tool for sampling nehu 

 larvae. 



Variance of the corrected length-frequency 

 distribution was used above to define some 

 length range that can be corrected for extrusion 

 and avoidance with sufficient precision, but esti- 

 mates of variance have other important uses, 



453 



