Boggs: Bioenergetics and growth of Engraulis mordax 



557 



Table 1 



Summary of feeding levels in sequence of application, 12-hour enforced speeds, average fish mass, and estimated daily rations for 

 the six treatments using the northern anchovy Engraulis mordax. N c = initial number of fish in each flume section at the start of 

 each treatment, N = total number killed for samples (number of samples in parentheses), and D = percentage of fish that died. 



Ration energy 

 (cal-g fish wet mass^'-day 1 ) 



55.7 

 52.0 





 



37.6 

 37.2 



* Average from daily amounts dispensed and equations in Figure 2. 

 ** Corrected for gastric evacuation (cf. Equation 4 in text). 



to prevent food in one section from passing to the 

 other. Uneaten food collected on the screens was 

 weighed. 



Rations were chosen to match the high ration (4-5%) 

 estimated by Hunter and Leong (1981), and a lower ra- 

 tion (3%) in the middle of the range of field estimates 

 (Blaxter and Hunter 1982). The zero ration was in- 

 cluded so that metabolism could be estimated from the 

 loss of body energy. 



Preliminary estimates of rations were made from the 

 net mass of food dispensed and the number of fish in 

 each treatment (Table 1). Preliminary ration estimates 

 were calibrated based on the amount of food in stom- 

 achs after feeding, corrected for gastric evacuation. To 

 estimate the gastric evacuation rate, 15-24 fish were 

 killed and stomach contents weighed once per hour for 

 7 hours on the seventh day of the first treatment. 



Removing and counting fish 



A sample of fish (Table 2)* was collected in one sweep 

 of a large dipnet at the beginning and end of each treat- 

 ment, and every 4-7 days during five of the six 

 treatments. In the low-ration, fast-speed treatment 

 there were enough fish only for the initial and final 

 samples. At least 20 fish were planned for each sam- 

 ple, but the number of fish netted was sometimes 

 lower, or much higher (Table 2)* because the sample 

 was taken quickly to avoid disturbing the fish. 



The standard length of sampled fish, their wet mass, 

 and the wet mass of stomach contents were measured. 



* In the zero-ration treatments, sample size was the same as the 

 number dried (Table 3). 



A subset of about 20 fish from each sample and the 

 combined stomach contents from 9 samples were dried 

 for 6 days at 55°C, cooled in desiccators, and weighed. 

 A census was kept on the number of fish in each sec- 

 tion of the flume based on rapid counts of the fish 

 transferred to the flume, minus removals due to sam- 

 pling and mortality. Dead fish were collected and 

 measured each day. The initial counts were accurate 

 to within 10% of the total number collected at the end 

 of the treatments. A corrected census, figured from 

 the total counts, was used in the final calculations. 



Data analysis 



Within each treatment, trends in sample means (i.e., 

 length) or calculated values (i.e., kcal/fish) were ex- 

 pressed as slopes in relation to elapsed time by using 

 linear regression (method of least squares). When only 

 two samples were collected in a treatment, the dif- 

 ference over time was calculated. The significance of 

 slopes, differences between initial and final sample 

 means, and differences between values estimated from 

 regression (i.e., log of dry mass) were evaluated with 

 t -tests. 



Apparent growth trends in some treatments were 

 corrected to account for high mortality of fish smaller 

 than the treatment mean. If there was no significant 

 change in length (e.g., in zero-ration treatments), then 

 no correction for mortality was made and length was 

 assumed to be constant. An approximation of the bias 

 in group mean length caused by mortality each day was 

 calculated as the difference between the apparent mean 

 length of the group each day (L A ) and the mean 

 length calculated by including the dead fish collected 

 that day. 



