Data Preparation Results & Discussion: Morisita's Index, dbh Distribution Analysis & Significant Variable Identification

The data analysis's were either run on R or in analyzed in Excel.

Data Preparation Results & Discussion: Determining Optimal Subplot Size to Allow for a Realistic Clustering Coefficient

Analyzing sizes from 10mx10m up to 40mx40m determined the optimal size for the new subplots was to be 30mx30m. Subplots that were smaller then the optimal size did not capture the clustering that was occurring and subplots that were larger then the optimal size captured greater portions of the surrounding area which data was not available for.

Data Preparation Results & Discussion: Determining Clustering Value for Each Species in Each Stand

The values calculated by the Morisita's Index can be viewed in the Cluster Data Table. The minimum frequency constraint was lifted when calculating a clustering value for species which had high frequency in 2 of the three stands, since in all cases the clustering value was calculated in the low frequency stand equaled 1, indicating the species was located randomly within that stand or 0, indicating the species were located uniformly within the stand. If the frequency of the species in the stand was 1 or 0 a clustering value of 1 was also assigned to the species in that stand.

Concern arised when using the Morisita's Index since the number of subplots in each of the stands was not equal. Since the Old Growth stand is almost 3 times the size of the Young and Mid Growth stands there are significantly more subplots in the Old Growth stand then the other research stands. This can be a problem since the Morisita's Index accounts for the number of subplots being sampled and as a result clustering values in the Old Growth stand would be larger since they are being multiplied by a larger n (number of subplots) value. While the number of subplots is larger in the Old Growth stand values of clustering are not compared between stands, but rather within stands. Although the number of subplots differs within a stand the relationship of clustering amongst a species remains constant, therefore the number of subplots within a stand is irrelevant in this analysis.

Data Preparation Results & Discussion: Determining Slope of the dbh Distribution of Each Species in Each Stand

The slopes of the dbh distributions for early and late succession species in each stand follow the outlined criteria well, but the variation in the calculated slopes for mid succession species was greater in the 3 stands. Overlapping in regeneration times and variation in growth characteristics can account for this discrepancy.

Select the stand title for graphic examples of species within the stand and how the slopes were calculated.

PLOTA -Young Growth Stand

PLOTB -Mid Growth Stand

PLOTC -Late Growth Stand

The slope values calculated by the analysis of the dbh distributions can be viewed in the Slope Data Table.

Data Results & Discussion: Determining Significant Variables for Multivariate Analyses

From previous research 5 early, 8 mid and 9 late successional species were classified using seed size, flower size and other traditional variables.

The resulting comparison mean table is:

 FreqA ClustA dbhA FreqB ClustB dbhB Freqc ClustC dbhC Early 42.25 2.70358 -0.93201325 8.75 0.7261905 N/A 32.75 4.396491 0.18637075 Mid 17 3.802896875 -0.20248825 47.125 1.472673375 -0.0723725 52.75 2.104919475 -0.482347 Late 11.66666667 1.250560222 -0.013661333 14.22222222 0.3076081 -0.392445 64.11111111 1.736597767 -0.080582889

The mean slope for the dbh distribution for the early successional in the Mid Growth stand was unable to be calculated because the frequency of the 4 species in the Mid Growth stand was less then 5 individuals.

Based on the means, the variables that describe both early and late successional species fit the hypothesized conditions and therefore those variables were included in the multivariate analyses. The variables that describe the mid successional species fit the hypothesized conditions except for the dbh slope in the Mid Growth stand (PLOTB). therefore the slope of the dbh distribution in the Mid Growth stand (PLOTB) was dropped from the data set. Comparing the correlations amongst the variables also resulted in the same conclusion. By dropping this variable 10 additional species were added to the test data bringing the total number of species for the multivariate analyses to 35 species which included 5 early, 4 mid and 5 late species whose succession classification was all ready known. Each of the 35 species had a complete data set therefore the multivariate analyses could be completed.

Slecet the Data Table link to view the final complete data set that was used in the multivariate analyses.

HOME
INTRODUCTION
DATA DETAILS
MULTIVARIATE METHODS
MULTIVARIATE RESULTS & DISCUSSION
CONCLUSION
APPENDICES
REFERENCES & ACKNOWLEDGEMENTS

DATA PREPARATION METHODS
DATA PREPARATION RESULTS & DISCUSSION

PRELIMINARY ANALYSIS