44 TH E STUDY OF PLANT COMMUNITIES * Chapter III 



to form the desired radius. The rod was thrust into the ground at 

 the sampling point and rotated so that the stick marked the limits 

 of the circle to be sampled. He said, "The most convenient forms 

 and sizes of the unit areas are the best!' With low vegetation, 

 circular plots are sometimes a distinct advantage. An efficient 

 means of laying out circular plots is to use a set of hoops or rings 

 of proper size tossed in all directions from a central point. These 

 cannot be used in tall vegetation of any kind since they may be 

 obstructed when thrown or may be suspended above the ground. 

 Larger circular plots can be quickly and accurately marked with a 

 string attached to a free-turning ring on a central axis. Again, this 

 method will not be found satisfactory where vegetation is more 

 than waist high. 



It has been demonstrated that a rectangular plot is significantly 

 more efficient in sampling than a square one of equal area since it 

 will tend to include a better representation of the variation in the 

 stand. Clapham, 55 who worked with low herbaceous vegetation in 

 his study of this problem, concluded that plots l A x 4m. were the 

 most efficient in size and that to secure the same amount of infor- 

 mation with squares as with strips nearly twice as large an area 

 would have to be observed. Short strips (1:4) gave less variable 

 data than squares but more variable than long strips (1:16). The 

 same general conclusions were reached after studies of certain 

 types of sagebrush-grass range sampling. 194 



Size and Number of Quadrats —A. community is rarely homo* 

 geneous throughout as to species and their distribution. Newly 

 formed habitats, such as sandbars or tidal flats where often only a 

 single species is a pioneer, may support a nearly homogeneous 

 stand, but the usual community will have some variation. If there 

 were no variation, a single relatively small sample would always be 

 sufficient. Since variation is the rule, it becomes necessary to have 

 samples large enough or numerous enough to include the variation 

 and to have it fairly represented in the data. There is thus always 

 a question of how large and how numerous the quadrats should be 

 for adequate sampling. 



The literature dealing with this problem is far too extensive to 

 review here. Agreement has not been reached on all phases of sam- 

 pling methods, and probably different methods will be advocated 



