BULLETIN FLORIDA STATE MUSEUM


constructed on each site, but this proved to be impractical due to the time required in
maintenance of the traps, as well as the destruction of traps by humans and cows. Con-
sequently, reptiles and amphibians were simply noted when observed in the course of
other field work. The result was a list of species without quantification of relative abun-
dance.
 LARGE MAMMALS.-Much the same situation existed with this class of wildlife as with
the herpetofauna. A systematic, quantified method of counting tracks along dirt roads was
applied but proved unfeasible because of the clay substrate and heavy vehicular travel.
Tracks, scats, and sightings were recorded when encountered during other activities on
the sites.

 DATA ANALYSIS

 The simplest method of characterizing post-mining treatments was by comparing lists
of the kinds of plants and animals that were present after recovery. However, more rig-
orous comparisons involved indices of abundance and diversity that reflected differences
among the communities supported by the post-mining treatments. The number of indi-
viduals and biomass provided measures of relative abundance. These two variables nor-
mally show similar trends for animal communities, but not for plants. The Shannon-Wie-
ner index (Shannon and Weaver 1949) was used to measure species diversity, defined as
H' = Yp, In pi/p,, where p, is the proportion of individuals in species i. This diversity
index has two major components. The first is the number of species, or species richness,
which is the fourth type of variable used in our comparisons. The second component is
equitability, or the evenness of distribution of the species, and is defined as E = H'/H,,,
(Sheldon 1969), where H._ is the natural log of the number of species.
Equitability was not extensively used in this report. Like the measures of abundance, H'
and the species number usually show similar trends of diversity (for both plants and ani-
mals). Generally all four quantitative evaluations are presented together to provide a thor-
ough view of community structure and to emphasize the differences and similarities be-
tween treatments.
 STATISTICAL ANALYSIS

 Statistical analyses of these data were performed by version 79.2B of the Statistical
Analysis System, using the computational facilities of the Northeast Regional Data Center
at the University of Florida.
 For statistical analysis the results were separated into two data sets-the annual or
seasonal measurements of 12 variables involving plants, herpetofauna, and large mam-
mals, and the seasonal (quarterly) measurements of 9 variables involving small mammals,
birds, and insects. Two sets of hypotheses were posed for each data set. One tested for
differences among the relatively stable end results of post-mining land use, including five
treatments (consolidated clay settling areas, unreclaimed pits and spoil piles) > 30 years
with lakes and without lakes, and both ungrazed and grazed pastures). The other tested
for trends over the years of the successional sequence on unreclaimed pits and spoil piles
with lakes. Rather than evaluating seral sites simply as replicated sites within treatments,
each was considered as an observation with a specific age in years. Throughout the anal-
ysis, some data were transformed to ensure constant variances (Steel and Torrie 1960).
Small value counts and measurements of area were transformed by (y + 0.5)1/2. Large
value counts were transformed by log (y + 1). And weights were transformed by y1'3. F-
tests were made with alpha = 0.05 using multivariate analysis of variance. Because of
sites and variables missed during autumn sampling, the autumn data were deleted from
the seasonal data set. For both site age and final land uses, analysis of variance was begun
by testing for interaction between season and age or treatment. No significant interaction


VOL. 30 NO. 3