D 



August 1957 



GRAVEL BAR 



August 1958 



Figure 8.--Dissolved oxygen levels at study area C (13-foot tide 

 level of Indian Creek). Samples were obtained 7 to 10 inches 

 beneatti gravel surface at points shown. 



Figure 9.--Dissolved oxygen levels at study area D (IT-foottide 

 level of Indian Creek). Samples were obtained 7 to 10 inches 

 beneath gravel surface at points shown. 



schemes are sufficiently precise to detect 

 significant differences in oxygen levels in 

 spawning gravels. For routine evaluation of 

 oxygen levels, there are certain advantages 

 to sampling randomly. They are: 



1. Sampling areas may be of any 



2. To obtain tiniformly precise estimates 

 of mean dissolved oxygen levels at any time, 

 sampling effort may be equally allocated 

 among areas, regardless of their size. 



3. The sampling effort required to obtain 

 a fairly precise estimate of niean dissolved 

 oxygen level of intragravel water is not 

 excessive. 



4. Changes in dissolved oxygen levels with 

 time may be determined by sampling individual 

 areas on tw^o or more occasions. 



5. Spatial differences in dissolved oxygen 

 levels may be determined by sampling two or 

 more areas simultaneously. 



With regard to points 1 and 2 above, ex- 

 amination of data given in the appendix indi- 

 cates that temporal and spatial variations are 

 of a similar magnitude in most spawning 

 riffles. It is therefore possible by sampling 

 equal numbers of random points to estimate 

 the mean dissolved oxygen content for a 

 stream or a single riffle with almost equal 

 precision. 



With regard to point 3, it has been observed 

 that the greatest variations in dissolved oxy- 

 gen levels occur in late summer during and 

 after spawning. By sampling 100 random points 

 at this time, the expected 95-percent confi- 

 dence limits of the mean dissolved oxygen 

 content of intragravel water is approximately 

 tO.5 mg./l. of the sample mean. At other 

 times, the mean dissolved oxygen level can 

 be estimated with almost equal precision by 

 sampling 50 points. 



11 



