arguably the most important difference between 

 the nets used by the programs is that the 

 FSFRL 2 m net has 1.7 mm mesh in the forward 

 section. Larvae are apparently lost by mesh 

 escapement through this section. This type of 

 multimesh construction may "herd" the larvae, 

 but this advantage seems to be greatly out- 

 weighed by escapement. If we assume that the 

 1.7 mm mesh retains no larvae, then catches by 

 the 2 m net can be recalculated for the 1.05 m 

 diameter opening of the remaining 0.5 mm mesh 

 section. This gave mean catches of 12.3 and 5.09 

 larvae/1,000 m^ for surface and oblique tows, 

 respectively. Based on this assumption, the 

 CSIRO catches were still 12 times larger than 

 FSFRL catches for surface tows and about 3 

 times greater for oblique tows. Reanalysis of the 

 ANOVA using these data produced the same 

 differences and levels of significance as in the 

 original analysis. This demonstrates that, al- 

 though escapement through the 1.7 mm mesh 

 may account for a significant loss of larvae, other 

 factors must also contribute to the lower catches 

 by the FSFRL progi-am. 



A number of factors could cause increased 

 avoidance of the 2 m net and hence reduce 

 catches. While it might be expected that the 2 m 

 net would reduce net avoidance by virtue of its 

 size, this may have the opposite effect owing to 

 an increase in the reaction distance to it (Barkley 

 1964, 1972; Wiebe et al. 1982). It has been well 

 established that samplers that have no bridle 

 obstructing the mouth (such as bongo nets and 

 the 70 cm net) are far more efficient samplers 

 than nets with conventional triple-rigged bridles 

 (such as the 2 m net) owing to increased avoid- 

 ance of the latter (Posgay and Marak 1980). 

 Other factors that could increase avoidance of 

 the 2 m net include the 30 kg depressor attached 

 directly to the frame of the net, increasing visi- 

 bility and turbulence and hence the reaction dis- 

 tance, and the high visibility of the undyed 

 (white) net compared with the blue 70 cm net 

 (Smith and Richardson 1977). 



The volume filtered by the 2 m net may also 

 have been overestimated as it was measured 

 outside the net by the depth/distance recorder. 

 This would result in lower apparent catch rates. 

 Reduced filtering efficiency due to clogging in 

 the 0.5 mm section, which may be a problem 

 with the 2 m net because of its low open area 

 ratio (Tranter and Heron 1967), would not be 

 detected. In the laboratory, the sorting of larvae 

 from samples without the aid of a microscope 

 could result in extraction of fewer larvae in the 



FSFRL program, thus further reducing appar- 

 ent catches. 



The 70 cm net caught greater numbers of tuna 

 larvae in the surface tow relative to the oblique 

 tow than did the 2 m net. Catches in the 2 m net 

 surface tow may have been reduced by increased 

 avoidance because the net is towed very close to 

 the hull of the Shoyo Maru. Alternatively, the 

 oblique tow may have oversampled surface 

 waters where the larvae are more abundant. The 

 tow profiles for the 2 m oblique tow do suggest 

 that this occurred to some extent. Clogging of 

 the 0.5 mm mesh in the surface tows would not 

 account for the disproportionately low catches 

 because clogging would have been more likely to 

 occur in the oblique tows which caught greater 

 volumes of plankton. 



We are not sure of the reasons for the differ- 

 ences in lengths of larvae caught by the two 

 programs because of the many possible con- 

 founding factors. The gi'eater mean length of 

 larvae in the FSFRL obhque tows was due to 

 fewer small larvae rather than to greater catches 

 of large larvae. This could have resulted from 

 differences in net deployment or in processing of 

 samples, or could simply have been an artifact of 

 the small sample size collected by the 2 m net in 

 oblique tows. It would not have been due to 

 differential loss of small larvae through the 1.7 

 mm mesh because this did not occur in surface 

 tows with the same net. 



The estimated abundance of tuna larvae in 

 CSIRO surface and oblique tows was much 

 higher than the corresponding estimates by the 

 FSFRL program. We assume that larger 

 catches are more accurate than smaller catches 

 because towed nets are inefficient samplers 

 owing to avoidance and mesh escapement (Clut- 

 ter and Anraku 1968; Murphy and Clutter 1972; 

 Clarke 1983) and subsequent treatment of sam- 

 ples (removal from nets, sorting, and enumera- 

 tion) is hkely to lead to the loss of larvae result- 

 ing in underestimation rather than overestima- 

 tion. This would suggest that the 70 cm net as it 

 is deployed in the CSIRO program is more effi- 

 cient at catching tuna larvae than the 2 m net 

 used in the FSFRL program, despite its much 

 smaller size. 



Acknowledgments 



We would like to thank the Masters and crews 

 of FRV Soela and FRV Shoyo Mam. for the 

 co-ordinated sampling at sea. Sorting, identifica- 

 tion, and measurement of fish larvae from 



980 



