PROBLEM SET #1

PBIO-415/515 (2004)

 

UNIVARIATE ANALYSIS

 

 

 

Background

 

Thompson et al. (2002; Ecol. Appl. 12: 1344–1363) examined the effects of historical land use on tropical forest in an effort to understand present forest characteristics and to plan conservation strategies. They compared the effects of past land use, topography, soil type, and other environmental variables on tree species composition in a subtropical wet forest in the Luquillo Mountains, Puerto Rico. The study involved stems ³ 10 cm diameter measured at 130 cm above the ground, within the 16-ha Luquillo Forest Dynamics Plot (LFDP), and represents the forest at the time Hurricane Hugo struck in 1989. Topography in the plot is rugged, and soils are variable. Historical documents and local residents described past land uses such as clear-felling and selective logging followed by farming, fruit and coffee production, and timber stand improvement in the forest area that now includes the LFDP. These uses ceased 40–60 yr before the study, but their impacts could be differentiated by percent canopy cover seen in aerial photographs from 1936. Using these photographs, they defined four historic cover classes within the LFDP. These ranged from cover class 1, the least tree-covered area in 1936, to cover class 4, with the least intensive historic land use (selective logging and timber stand improvement). In 1989, cover class 1 had the lowest stem density and proportion of large stems, whereas cover class 4 had the highest basal area, species richness, and number of rare and endemic species. Ordination of tree species composition (89 species, 13167 stems) produced arrays that primarily corresponded to the four cover classes (i.e., historic land uses). The ordination arrays corresponded secondarily to soil characteristics and topography. Natural disturbances (hurricanes, landslides, and local treefalls) affected tree composition, but these effects did not correlate with the major patterns of species distributions on the plot. Thus, it appears that forest development and natural disturbance have not masked the effects of historical land use in this tropical forest, and that past land use was the major influence on the patterns of tree composition in the plot in 1989. The least disturbed stand harbors more rare and endemic species, and such stands should be protected.

 

 

Analysis

 

Thompson et al. (2002) provide an online appendix of their tree data available through the Ecological Society of America Archives. Download the data set. Bring it in to Excel and clean it up for analysis. We are interested only in the LFDP density data (labeled “No. stems per ha”) and the size class data (labeled “Typical mature ht.”: Sm, M, L)—all other data are extraneous for this analysis and can be deleted.

 

First, sort the data into the 3 size classes. NB: some species are dual listed (e.g., Sm/M); for the purposes of this exercise, assign half of the density to each size class. Sort and summarize the data by size class, produce histograms, box plots, normal probability plots, etc. Evaluate the distribution of each data set for symmetry, kurtosis, modality, outliers, etc. (both quantitatively and qualitatively).  Quantitatively describe the central tendency using the mean, median, and mode.  Evaluate the spread of the data via estimates of the range, variance, standard deviation, percentile distributions, etc.

 

Provide a short expository explanation of your data, the summary stats and graphics, and discuss the possible implications of your findings.