passes from one type of terrain or vegetation to an- 

 other. 



A variation of this method, often employed for 

 counting larger animals such as deer, is to increase 

 the width of the census strip by using a line of many 

 observers that progresses uniformly over an area of 

 previously fixed dimension. The animals are counted 

 as they are driven back through the line or out be- 

 tween other observers stationed along the boundary 

 (Rasmussen and Doman 1943). Helicopters may be 

 efifectively used for counting large animals in open 

 country (Aldous 1956) ; faster flying aircraft are less 

 successful (Gilbert and Grieb 1957). 



SAMPLE PLOTS 



Since it is seldom possible to count all the 

 individuals present in a large area, it becomes neces- 

 sary to take sample counts over small areas where 

 accurate counting of individuals is practical. The 

 problem then arises as to the number, size, shape, and 

 distribution of plots required to give reliable infor- 

 mation on species composition and the mean density 

 for all the organisms involved. Much work on this 

 problem has been done by plant ecologists, and their 

 techniques should also be of use to animal ecologists. 



Plot distribution and shape 



Sample plots may be distributed either sys- 

 tematically or at random. Systematic arrangement of 

 plots of uniform size spaced at equal intervals along 

 straight lines is often preferred because of its easy 

 application. However, if the distribution of organ- 

 isms over the area shows a uniform pattern of vari- 

 ation, systematic sampling may indicate densities 

 either too high or too low. Furthermore, systematic 

 sampling does not permit the assessment of error, 

 since statistical theory requires that the location of 

 each sampling unit be independently determined, 

 whereas in systematic sampling, the position of all 

 plots is determined by the location of the first one. 

 The completely random location of sample plots over 

 an area may be somewhat more difficult to apply in 

 the field, but the data obtained are just as precise 

 and have the advantage that the error of sampling 

 can be calculated (Bourdeau 1953). In order to get 

 randomly located sample plots, a map of the entire 

 area is subdivided into numbered plots of the proper 

 size. The plots to be used are then selected by using 

 tables of random numbers. If the same number comes 

 up twice, the duplication should be discarded (Dice 

 1952). Other plans of sampling, such as stratified 

 random sampling, may sometimes be preferable. 



Where a habitat is perfectly uniform, the shape 

 of a sample plot is not of great importance, although 

 square plots are commonly used. Where a habitat is 

 obviously not uniform a rectangular plot oriented with 

 its long axis across any observed contour-, soil-, or 

 vegetation banding will furnish less variable data 

 than plots that are shorter and wider (Bormann 

 1953). Circular plots, which possess a smaller 

 periphery than any other shape, are useful where the 

 influx and exit of animals must be minimized. 



The size of plot suf^cient to include an ade- 

 quate sampling of the species composition of a par- 

 ticular local community varies with species involved 

 and density of populations. Larger plots must be 

 used for larger organisms, richer fauna, for situations 

 in which one or a few species are so markedly pre- 

 dominant that minor species are scattered, and where 

 population levels generally are low. Since the number 

 of species included will vary with the size of the area 

 covered in sampling, some standardization is desir- 

 able for comparing the species composition of dififer- 

 ent communities. 



A standard size for sampling plots may be deter- 

 mined empirically (Vestal 1949). If the numbers of 

 species found on plots of different sizes are plotted 

 against the logarithms of the plot sizes, a sigmoid so- 

 called species-area curve is formed. The characteris- 

 tics of this curve are that an increase in the size of 

 small sampling plots includes, at first, a considerably 

 larger number of species, but later a size of plot is 

 reached, varying with the kind of organisms being 

 counted, beyond which there is little to be gained by 

 increasing the area sampled. Two arbitrarily chosen 

 points on the upper part of this curve, where it is 

 concave toward the scale of plot size, have ecological 

 significance. One of these points represents a plot 

 fifty times the size of the other, containing twice the 

 number of species of the other. The larger plot is 

 close to the upper asymptote of the curve and repre- 

 sents a fair-sized sample plot for practically all pur- 

 poses. The smaller area, located near the point of 

 inflection and containing half the number of species, 

 is the smallest representative area that is sufficient to 

 identify the community, but hardly usable for any 

 other purpose. A third point may be identified, mid- 

 way between these two points on the curve, as the 

 minimum area large enough to include all important 

 species and about half of the minor ones. It clearly 

 defines the community and the approximate ranking 

 of species in points of number and biomass. The area 

 this intermediate point represents is five times the 

 smallest representative area and one-tenth the fair- 

 sized area. 



32 Background 



