1987 SCHNOES AND HUMPHREY: TERRESTRIAL COMMUNITIES IN FLORIDA 101

 STATISTICAL ANALYSIS

 The results of tests of hypotheses on contrasted final land uses are
presented in Table 13. Considering annual data, the shrub and tree var-
iables showed significant differences among end results of treatments,
whereas herb, herpetofaunal, and large mammal variables did not. Dun-
can's multiple range tests of end results of reclamation showed two sig-
nificantly different groups->30 year old unreclaimed sites with and
without lakes had high mean values, and ungrazed and grazed reclaimed
pastures had low values. The clay waste treatment fell between these
groups with intermediate values but was not clearly assignable to either.
The two reclaimed treatments were never significantly different. In cases
where the clay treatment moved up in mean value to second or first
place, it was never significantly different from the third place treatment.
When the two unreclaimed treatments were significantly different, the
treatment with lakes had the higher mean value. For variables with this
relation reversed, the differences were not significant. The contrasts re-
inforced these observations without showing additional distinctions.
 Considering seasonal data (Table 13), only bird variables showed sig-
nificant differences among final land uses; insect and small mammal var-
iables did not. Response of seasonal variables to treatment end results
was weaker than their successional responses, for distinct reasons in each
group of animals. For small mammal data, no seasonal effect appeared in
the analysis of variance, but the effect of sites within treatments was
stronger than the overall treatment effect. Hence high site variance pre-
vented distinction of post-mining treatments (Table 13) with small mam-
mal data. By comparison, the strongest effect on bird variables was the
season effect. The Duncan's grouping of final land uses and the contrasts
showed patterns similar to those for annual data, with two different fea-
tures. The clay waste treatment was never significantly different from
the reclaimed pastures. Ungrazed pastures were significantly higher
than grazed pastures in bird H'.
 The results of hypotheses on response to site age also are presented
in Table 13. Considering annual data, the shrub and tree variables
showed significant responses to age, whereas herpetofaunal and large
mammal variables did not. Considering seasonal data, small mammal and
bird variables showed significant responses to age over the interval 0-36
years, and insect variables did not. Tests showed no simple pattern of X2,
X3, or X4 functions for all variables.
 Models of plant and animal response to age, based on statistically sig-
nificant age effects (Table 13), yield predictions of community succession
on unreclaimed overburden soil (Fig. 21). Shrub species colonized for
the first 9 years after mining (Fig. 21A); then the number of species