APPLICATION OF PHYTOSOCIOLOGICAL TECHNIQUES 295 



ured trees) 153 species among 897 trees with the following distribution: 

 1 specimen only, 67 species; 2 specimens, 24 species; 3 specimens, 13 species, 

 and so on: 4—9, 5—3, 6—3, 7—4, 8—5, 9—4, 10—1, 11—4, 12—1, 13—1, 

 14—2, 16—1, 17—2, 21—1, 22—1, 26—1, 27—1, 39—1, 49—1, 58—1, 

 69 — 1, and 101 stems, 1 species. Preston (1948) concluded that random 

 samples of ecological assemblages indicate that the universes from which they 

 are drawn have approximately the form of an ordinary Gaussian curve drawn 

 upon a logarithmic base. Samples have the same form as the universe but 

 are truncated at the left, a rather large number of rare species not being 

 sampled. Preston approaches the problem by determining the number of 

 species in each of a series of frequency classes of logarithmic nature, or oc- 

 taves, as in the natural series of commonness: 1, 2, 4, 8, 16, 32, 64. . . . The 

 octave 4-8, for example, would contain all the species represented by 5, 6, 

 and 7 individuals in a sample, and half the species represented by 4 and 8 

 individuals. For our Mucambo study the octaves have the following values: 

 <1 — ^^Y2 (i-C, half the 67 species represented by a single specimen on the 

 plot); 1-2 — 45^/4 species (half of the singles and half of the doubles); 2-4 

 — 29^/^ species (half the doubles, all the species represented by three plants, 

 and half those with four plants). Similarly, the remaining octaves are: 4-8 

 —17 species; 8-16 — 16 species; 16-32 — 6^2 species; 32-64 — 3 species; and 

 64-128 — 2 species. These data are plotted in fig. 2. It is clear the model 

 class is that of 1-2 specimens on this 2-hectare sample plot. This curve 

 indicates that the total flora of the vicinity of this plot and within the parti- 

 cular forest type is probably in the neighborhood of 200 species of trees with 

 diameters of 10 cm. or more. In other words, the 2-hectare plot sampled 

 about three-fourths of the trees as defined here. Since we do not have density 

 data for smaller trees, shrubs, etc., it is not possible to make a similar esti- 

 mate of the total flora of the community. 



Our data on frequency and basal area as a measure of dominance do not 

 seem to need discussion. These and other phytosociological concepts are 

 discussed in Braun-Blanquet (1932), Cain (1932), Costing (1948) and in 

 an extensive journal literature. The idea of combining data by the use of 

 percentages, as in relative density, relative dominance, and relative frequency 

 to form a sum known as the importance value index originated in the work 

 of Curtis and his colleagues. Curtis and Mcintosh (1951) modified the 

 original method in a paper on the upland forest continuum in southwestern 

 Wisconsin. They claim that this index is an excellent indication of the 

 vegetational importance of a species within a stand. In temperate forests 

 it is possible to estimate rather accurately the relative importance of the few 

 species of trees, but the index permits a measurement and is a good device 

 for comparisons of different stands. For the herbaceous layer of northern 

 forests such estimation is difficult and the objective determination of the 

 importance value index is an important device. In equatorial rain forests 



