possible, the use of hand-operated core samplers obviates most of the 

 problems associated with large shipboard grab-sampling devices (e.g., 

 inconsistent sampler volume, penetration depth, and disturbance of 

 sediment). The compromise made is one of smaller sample volume of 

 corers and the general physical limitations of working underwater. 

 The advantages are that greater replication is possible and samples 

 may be spaced according to design and taken with little disturbance of 

 surface sediments. This is important since most small infaunal organisms 

 reside within the upper 4 to 5 centimeters in benthic sediments. 



The use of different sampling devices as well as a wide variety 

 of screen sizes for sorting organisms has made it difficult to compare 

 data between studies (Reish, 1959; Knox, 1977). In particular, data 

 obtained for species diversity and abundance are highly contingent on 

 screen size. Selection of screen size is based on sediment type and a 

 balance between the specific type of information desired (e.g., diversity, 

 biomass) , as well as time limitations for sorting samples and making 

 taxonomic identifications. Experience with nearshore samples from 

 differing sediment types in southern California has indicated a twofold 

 to threefold increase in sorting time between 1.0- and 0.5 -millimeter 

 screen sizes. Recently, monitoring programs have used smaller screens 

 more frequently. A recommended strategy is to sieve samples through 

 nested screens (e.g., 1.0 millimeter above a 0.5-millimeter screen). 

 Material is preserved separately. Organisms from the larger screen 

 can be analyzed, then if cost or time limitations permit, a greater 

 level of description of the biota can be obtained by analyzing mate- 

 rial from smaller screen sizes. 



Since the time required to process samples and identify the or- 

 ganisms imposes a limitation on the number of samples which can be 

 processed within the constraints of any sampling program, choices deal- 

 ing with the timing, spacing, and number of replicates assume an added 

 importance. Sampling design should maximize information return within 

 these constraints. Optimization approaches have seldom been invoked in 

 benthic sampling programs. Recently, Saila, Pikanowksi, and Vaughan 

 (1976), Cox (1976), Scherba and Gallucci (1976), and Diener and Parr 

 (1977) have dealt with maximization of information return and problems 

 of precision in estimating sedimentary macrofaunal and meiofaunal popu- 

 lations. 



Analyses of data from benthic programs have traditionally utilized 

 parametric statistics to test for differences between populations. 

 However, assumptions underlying these analyses require preconditions 

 (equal sample variances, variance equal to the mean) which are not 

 usually met with biological data. More frequently, animals are aggre- 

 gated and their distribution quite often fits a negative binomial 

 distribution (Bliss and Fisher, 1953; Debauche, 1962) in which case 

 the variance exceeds the mean. Various approaches to analysis of 

 benthic data and transformations to normality for different types of 



19 



