Table 2. — Nonparametric sign test of catch sizes for selected 

 echinoderms in 22 pairs of samples. The long haul in each pair 

 was that with both wheel counts being greater than both wheel 

 counts of the other sample in the pair. When the counts over- 

 lapped, the pair was excluded since there was no unambiguous 

 longer or shorter haul. All sample pairs were taken at the same 

 locality on the same day. 



Criterion 



Highest value in 



Significance 



Long haul Short haul (PsO.05) 



Most echinoderm species 

 Most echinoderm specimens 



15 

 16 



ns 

 ns 



dant megafaunal holothuroids off Oregon at 

 2,500-3,000 m. The catches for each species were 

 compared with the wheel count estimates of dis- 

 tance towed (based on the average of the two 

 wheels) by computing correlation coefficients. Ac- 

 cording to wire-out determinations all of these 

 tows were on bottom 120 min. These coefficients 

 were computed with the zero holothuroid catches 

 included and then with zero catches excluded (Ta- 

 ble 3). There was no consistent pattern of positive 

 correlation. 



Table 3. — Correlation of number of specimens of each species of 

 holothuroid with distance on bottom sampled as determined by 

 wheel counts (see text). Correlation coefficients were computed 

 for each species over all samples (n = 100) and for only those 

 samples where the species was taken (zeros excluded) to reduce 

 the effects of possible aggregation. 



The earliest reported use of odometer wheels 

 was by Bieri and Bradshaw (cited in Gunther 

 1957) whose system evolved into a more sophisti- 

 cated opening and closing quantitative trawl 

 (Bieri and Tokioka 1968). Subsequently, Belyaev 

 and Sokolova (1960), Riedl (1961i, Gilat (1964), 

 Richards and Riley (1967i, Pequegnat et al. 

 (1970), and Carey and Heyamoto ( 1972) reported 

 the use of similar devices. Additionally, Wolff 

 ( 1961) presented a photograph attributed to Zen- 

 kevitch of a Soviet beam trawl carrying four 

 odometer wheels similar to those discussed in this 

 report. 



Positive correlation between counts, duration, 

 and position change is supportive of the basic con- 



tention that the wheels can provide a measure of 

 the distance sampled. However, major questions 

 remain as to the accuracy and precision of the 

 wheels, and the relationship between estimates of 

 area sampled and the catch results. It must be 

 stressed that the positive correlation among dis- 

 tance estimates does not mean that they are either 

 accurate or precise. Our regression of the limited 

 time-depth bottom time estimates on odometer 

 readings indicated that the wheels were inaccu- 

 rate, being biased by as much as 40^ below the 

 actual distance. The numerous low wheel count 

 ratios indicated that the wheels did produce rela- 

 tively low variance measurements most of the 

 time. 



Photographic evidence indicated that the lack of 

 a positive correlation between catch and wheel 

 count might have been due to irregularities that 

 affected total trawl performance, not just the 

 operation of the wheels. Using a camera mounted 

 to the trawl frame it was determined that during a 

 portion of each haul the trawl skids rocked for- 

 ward, lifting the footline of the net off bottom. The 

 odometer wheels, however, remained in contact 

 with the bottom registering the distance towed. 

 The severe saltations observed by Rowe and Men- 

 zies (1967) and Menzies et al. (1973) were not ob- 

 served in these photographs and did not appear to 

 be a problem with the beam trawl used in this 

 study. 



The failure of the footline to constantly tend 

 bottom does not detract from the usefulness of 

 odometer wheels, but it does make it difficult to 

 interpret faunal data in terms of density ( Pearcy 

 1978). Nevertheless, in some preliminary com- 

 parisons of trawl versus photographic determina- 

 tions of faunal densities good agreement has been 

 found. Pearcy (1972) measured populations of the 

 pink shrimp, Pandalus jordani, at shelf depths 

 using both techniques and got similar estimates of 

 about 10 individuals/ m^. Similarly Carey et al.^ 

 measured ophiuroid densities at about 2,500 m 

 and got similar estimates of 2 or 3 specimens /m'^. 



Conclusions 



The system of trawl frame and odometer wheels 

 used in this study did not produce estimates of 



^Carey, A. G., Jr., J. Rucker, and R. Tipper. 1973. Benthic 

 ecological studies of deepwater dumpsite G in the northeast 

 Pacific Ocean off the coast of Washington. In Proceedings of the 

 First Conference of the Environmental Effects of Explosives and 

 Explosions (May 30-31, 1973), p. 120-137. Nav Ord. Lab. Tech. 

 Rep. 73-32, N.O.R.D.A., Bay St. Louis, MS 39520. 



795 



