FISHERY BULLETIN: VOL. 74, NO. 4 



types of transect methods, and Saila and Gaucher 

 (1966), Russell (1972), and Loesch (1974) described 

 subtidal stratified random sampling methods. 



Transecting methods used in the soft substrate 

 of the intertidal zone usually involve the choice of 

 a narrow belt of one or two sampling units in 

 width, placed perpendicular to the water line. 

 Samples are collected at fixed and predetermined 

 tidal levels (e.g., every 10 m) to correspond to 

 changes in such things as algal and sediment 

 composition (Matthiessen 1960; Vassallo 1969; 

 Warme 1971; Bloom et al. 1972). This method will 

 be referred to as fixed level transect sampling 

 (FLTS). A common denominator in this type of 

 work is that no probability model is used in 

 selecting the location of the sampling units. An 

 alternative to FLTS is simple or stratified random 

 sampling where the discrete uniform probability 

 distribution underlies the selection of sample sites. 

 The disadvantage of random sampling is that 

 there is no guarantee that sample sites will be in 

 those areas where experimental interest is 

 focused. However, without an underlying 

 probability model, valid statistics may not be 

 estimated (Cochran 1963) because the sample sites 

 may not be independently located and subsequent 

 statistical tests may be invalid. These points are 

 often overlooked. 



Stratified systematic sampling (SSS) is 

 proposed as an alternative to the FLTS method 

 currently popular in intertidal fieldwork. The 

 usefulness of SSS is demonstrated by applying it 

 to a study of spatial and temporal variation in a 

 macrofauna-sediment relationship. This field 

 study was conducted at Garrison Bay (Figure 1), a 

 small embayment on San Juan Island, Wash. (lat. 

 48°35'N, long. 128°08'W). 



MATERIALS AND METHODS 



Applications of systematic sampling are found 

 in the forestry literature (Osborne 1942; Finney 

 1948; Matern 1960; Faber 1972). Mathematical 

 details are found in sampling texts such as Coch- 

 ran (1963), Raj (1968), or Sukhatme and Sukhatme 

 (1970) and in many theoretical papers.^ Systematic 

 sampling assumes that the sampling units in the 

 area to be studied are numbered consecutively. 

 The attractiveness of the method is increased 

 by the relatively sessile nature of many intertidal 



^Scherba, S., Jr., and V. F. Gallucci. 1976. Quantification of 

 species-presence gradients by stratified systematic sampling 

 and the autocovariance function. Unpubi. manuscr., 15 p. 



Figure 1. -Shoreline of Garrison Bay showing the location of the 

 four study areas. Insert shows the representative arrangement 

 of the subareas (A and B) and the strata (1 and 2) in these areas. 



organisms. From A^ sampling units numbered 1, 2, 

 . . ., N; n sampling units are selected, all evenly 

 spaced at a distance of K sampling units apart. 

 Thus, N = riK. The location of the first unit to 

 actually be sampled is randomly chosen by select- 

 ing a number between 1 and K from a table of 

 random numbers. Hence, systematic sampling is 

 based upon a uniform probability distribution (Raj 

 1968). SSS is a variation in which the region is 

 divided into strata, e.g., at the locations of the 

 fixed levels where samples would have been col- 

 lected using FLTS. Each stratum is independently 

 sampled in the manner described above. 



Four regions with different sediment composi- 

 tions were a priori defined in the intertidal area of 

 Garrison Bay (Figure 1). Representative areas 

 within these regions were sampled in the winter 

 (January and February), spring (May and June), 

 and summer (July and August) 1974. A north- 

 south sediment gradient exists because fine 

 sediment is deposited at the closed end of the bay 

 where the water is less turbulent. Thus, the south 

 (closed) end of the bay consists principally of fine 

 grades of sediment, while the north (open) end 

 consists mainly of coarser grades. Visual exami- 

 nation indicates that perpendicular to the water, 

 there is a sediment gradient as well as a zonation 

 of intertidal animals. However, the statistical 

 comparisons of the data from strata, which were 



938 



