DURBIN ET AL.; SWIMMING SPEEDS AND RESPIRATION RATES OF ATLANTIC MENHADEN 



brightwelli with an efficiency of about 25% (Dur- 

 bin and Durbin 1975). 



Respiration Rate 



Oxygen consumption by the fish was deter- 

 mined by closed system respirometry. The water in 

 the tank was sealed from contact with the atmo- 

 sphere by means of a circular cover made of clear 

 1.2 cm Plexiglas, suspended on pulleys over the 

 tank, which could be gently lowered onto the 

 water surface. Replicate water samples for oxygen 

 determinations (Strickland and Parsons 1972) 

 were siphoned from the tank through a sampling 

 port every 12 min during feeding measurements, 

 and every 20 min during nonfeeding mea- 

 surements. The precision of the method was 

 ±0.019 mg O2/I. Measurements of oxygen from 

 different locations in the tank demonstrated that 

 the movement of the fish kept it well mixed at all 

 times. Control measurements on the tank, filtered 

 seawater, and tank water after the addition of 

 phytoplankton demonstrated that these did not 

 contribute significantly to the change in oxygen 

 content of the water during respiration mea- 

 surements. The oxygen level in the tank was not 

 allowed to drop by more than 2 mg/1 during any 

 measurement; between measurements, the lid 

 was raised off the surface of the water, and air was 

 bubbled through the airstones along the tank 

 walls. The decline of oxygen in the tank with time 

 was linear, with a correlation coefficient of 0.98 or 

 better in all cases; the mean respiration rate of the 

 fish was calculated from the slope of this regres- 

 sion. Ninety-five percent confidence limits (CD 

 were computed for the slope and used to calculate 

 the 95% CL for the respiraton rate in each mea- 

 surement. Respiration rates are reported as mil- 

 ligrams oxygen consumption per gram wet weight 

 offish per hour (mg 02/g per h). 



Swimming Speed 



During the respiration measurements the 

 swimming behavior of the fish was recorded with a 

 Sankyo ES-44XL 8 mm movie camera, equipped 

 with a wide angle lens and a remote control and 

 mounted above the tank. The fish were photo- 

 graphed with Kodak Ektachrome 160 film, ex- 

 posed at 9 frames/s. Paired 10 s shots, 1 min apart, 

 were taken every 6-10 min while the fish were 

 feeding, and every 15-20 min when they were not 

 feeding. Films were later analyzed using a Kodak 



Model MPG-TH microfilm reader at a magnifica- 

 tion of 34x. A sheet of clear acetate was placed 

 over the viewing screen, and the location of each 

 fish was plotted at every fifth frame (the corres- 

 ponding time interval at 9 frames/s = 5/9 si when 

 the fish were feeding and swimming rapidly, and 

 every 10th frame (or, every 10/9 s) when the fish 

 were not feeding and swimming slowly. These 

 measurements were then converted to swimming 

 speed in centimeters per second and body lengths 

 (BL) per second. Vertical travel by the fish, which 

 was not corrected for in this method, was negligi- 

 ble since the fish tended to maintain themselves at 

 middepth in the water column. 



During each measurement of oxygen consump- 

 tion, an average of 680 observations of swimming 

 speed were obtained. The average swimming 

 speed during each measurement was determined 

 from the mean of all observations ±95% CL. The 

 distribution of swimming speeds within each mea- 

 surement was compared with a normal curve. 

 Measurements were compared by first testing for 

 homogeneity of variance, and when appropriate 

 the significance of the difference between means 

 was tested using analysis of variance. When var- 

 iances were nonhomogeneous, differences be- 

 tween means were tested according to a non- 

 parametric test. 



The mean values of each measurement were 

 used to determine the relationship between 

 swimming speed and respiration rate. Mea- 

 surements were grouped into three categories: ini- 

 tial and final (unfed), feeding, and postfeeding. In 

 the latter two categories there was a linear rela- 

 tionship between swimming speed and log oxygen 

 consumption. Predictive regressions of y on X are 

 presented to permit the comparison of present re- 

 sults with those in earlier studies. However, we 

 also present the functional regressions (GM) 

 (Ricker 1973), which represent the geometric 

 mean of the regression of Y" on X and the recip- 

 rocal of the regression of X on Y. Although there 

 has been some controversy on the subject 

 (Jolicoeur 1975; Ricker 1975) the functional re- 

 gression nevertheless appears to be the preferable 

 method of describing the data. Differences in the 

 slopes and elevations of the regressions were 

 tested for significance by covariance analysis. 



RESULTS 



The menhaden were 3 yr old, with a mean fork 

 length of 25.8 cm (range 23.0-27.9 cm), a mean wet 



879 



