OF OAK TREE LEAVES by
The group studied the (anticipated) relationship between the width and length of leaves, for two varieties of oak trees. It was expected that the ratio of width to length would be constant for a particular variety.
For one tree, data was categorized by the number of leaves in each cluster. The question arose as to the amount of variation among the clusters and the possible relationship of this to the "distance" between clusters.
Two oak trees, one a red oak and the other an unidentified variety of oak, on the Dickinson College campus, were chosen. From the red oak a "random" sample of leaves from reachable branches was selected. A wind-felled branch from the second oak constituted the other sample.
For each leaf in the sample, we recorded the length and width in millimeters. The length was measured along the stem vein from the tip to the base of the leaf (fleshy part of leaf), without consideration to curvature. The width was defined to be the widest distance between the tips of the lobes of a leaf running roughly perpendicular to the vein.
The collected data separates measurements by tree and by cluster of leaves. Cluster data was only collected from the unidentified variety of oak. Due to time restraints, the cluster information was not incorporated in the following report.
The following box plot demonstrates the variation within the samples for each of the trees and illustrates that the range of lengths appear to be different for the two trees.
Box plot of length
The following scatterplot indicates a possible linear relationship betwen width and length. It is apparent that the relationship is different for the two trees.
A summary of the basic statistics follow:
Tree # Sample Size Mean Median
length 1 42 13.5 13.
2 48 17.6 17.4
width 1 42 12.3 12.8
2 48 13.0 12.5
ratio 1 42 0.91 0.92
2 48 0.73 0.73
The Minitab display of the complete descriptive statistics are contained in the appendix.
Next , consider the regression information for each tree, separately.
For the red oak ( in scatterplot as tree #1, symbol black o)
The regression equation is width = 7.75E -02 + 0.904 length. The constant term indicates that the line passes through the origin as would be expected. The linear coefficient shows that the expected width is 0.904 times the length.The corresponding residual plot seems to show no noticable pattern, meaning that the linear regression line provides a satisfactory model.
For the second oak tree (scatter plot Tree#2, symbol +).
The regression line for tree #2 is width = -1.21 + 0.803 length. The linear coefficient shows that the expected width is 0.803 times the length. The corresponding residual plot seems to show no noticable pattern, indicating that the linear regression line provides a satisfactory model.
It appears that there is a strong relationship between length and width of leaves for the two oak trees and that this ratio was unique for the two varieties.
A large concern is the clearly non-random sampling methods used in this pilot study. A more complete study would consider leaf location - top vs. bottom, interior leaves vs. exterior leaves, sun vs shade. Also, trees from different geographical locations could be chosen -- rural vs. urban, open space vs. dense forest. The possible effect of tree age, tree health on the ratio of length to width could also be studied. This list is not exhaustive.
To study the relationship between the dimensions of the leaves a scatter plot and related regreeeion analysis was performed for the width (y) vs. length (x) for each of the trees.
The regression equation for tree 1 is: wid.1 = 0.077 + 0.904 len.1
The regression equation for tree 2 is: wid.2 = - 1.22 + 0.803 len.2
The regression equation is wid.1 = 0.077 + 0.904 len.1 Predictor Coef Stdev t-ratio p Constant 0.0775 0.8770 0.09 0.930 len.1 0.90444 0.06389 14.16 0.000 s = 1.045 R-sq = 83.4% R-sq(adj) = 82.9% Analysis of Variance SOURCE DF SS MS F p Regression 1 218.74 218.74 200.40 0.000 Error 40 43.66 1.09 Total 41 262.40
The regression equation is wid.2 = - 1.22 + 0.803 len.2 48 cases used 2 cases contain missing values Predictor Coef Stdev t-ratio p Constant -1.2156 0.8627 -1.41 0.166 len.2 0.80264 0.04817 16.66 0.000 s = 1.122 R-sq = 85.8% R-sq(adj) = 85.5% Analysis of Variance SOURCE DF SS MS F p Regression 1 349.29 349.29 277.62 0.000 Error 46 57.87 1.26 Total 47 407.16
Variable TREE.NO N NMiss Mean Median TrMean
length 1 42 0 13.493 13.600 13.558
2 48 2 17.592 17.400 17.616
width 1 42 0 12.281 12.800 12.326
2 50 0 12.950 12.450 12.925
ratio 1 42 0 0.9103 0.9239 0.9140
2 48 2 0.73159 0.73186 0.73273
Variable TREE.NO StDev SEMean Min Max Q1 Q3
length 1 2.554 0.394 7.900 17.600 11.550 15.925
2 3.396 0.490 9.800 25.200 15.175 19.775
width 1 2.530 0.390 7.000 16.600 10.325 14.200
2 2.908 0.411 7.100 19.300 10.650 15.450
ratio 1 0.0773 0.0119 0.6952 1.0462 0.8738 0.9553
2 0.06533 0.00943 0.57303 0.86624 0.68178 0.78643
length width stem lobe ratio cluster tree TREE.NO 14.0 13.3 3.1 7 0.95000 * 1 15.2 11.0 4.3 9 0.72368 1 16.0 15.6 5.4 7 0.97500 1 14.7 13.8 4.9 7 0.93878 1 11.1 11.5 2.5 7 1.03604 1 10.1 10.1 2.3 7 1.00000 1 8.8 8.8 2.4 7 1.00000 1 7.9 7.0 1.3 7 0.88608 1 14.3 13.1 4.5 7 0.91608 1 16.7 14.2 5.6 9 0.85030 1 16.3 12.7 5.1 7 0.77914 1 13.4 13.7 3.8 7 1.02239 1 17.2 16.0 6.5 7 0.93023 1 12.2 11.3 2.4 7 0.92623 1 10.0 8.6 2.4 7 0.86000 1 13.6 13.0 4.2 7 0.95588 1 10.4 8.6 2.2 7 0.82692 1 13.6 11.1 4.2 7 0.81618 1 12.8 12.2 3.5 7 0.95312 1 13.1 11.6 4.5 7 0.88550 1 13.0 13.6 4.3 7 1.04615 1 14.1 13.7 4.3 7 0.97163 1 15.9 14.2 5.3 7 0.89308 1 12.1 9.9 3.1 7 0.81818 1 12.5 10.4 3.4 7 0.83200 1 16.0 14.3 3.7 7 0.89375 1 12.3 11.4 2.9 7 0.92683 1 14.8 13.0 3.1 7 0.87838 1 16.6 16.6 3.4 7 1.00000 1 10.5 7.3 1.7 9 0.69524 1 10.2 9.3 1.9 7 0.91176 1 13.6 12.6 4.8 7 0.92647 1 15.6 14.6 4.6 7 0.93590 1 17.6 16.5 5.1 7 0.93750 1 16.0 15.9 5.2 7 0.99375 1 16.7 15.1 4.3 7 0.90419 1 15.6 14.9 4.9 7 0.95513 1 10.2 9.4 2.9 7 0.92157 1 9.6 8.7 1.9 7 0.90625 1 14.2 12.9 3.0 7 0.90845 1 11.7 10.9 2.8 7 0.93162 1 16.5 13.4 4.1 9 0.81212 1
length width stem lobe ratio cluster tree * 12.5 5.0 * G 2 12.4 9.8 3.5 0.79032 G 2 19.8 15.1 5.6 0.76263 G 2 15.7 11.0 4.8 0.70064 G 2 21.2 17.3 6.0 0.81604 G 2 17.1 13.0 5.0 0.76023 G 2 15.7 11.4 3.7 0.72611 E 2 19.2 14.1 9.1 0.73438 E 2 17.4 12.8 4.0 0.73563 E 2 16.7 13.6 4.7 0.81437 E 2 23.3 16.8 5.2 0.72103 E 2 20.6 16.8 5.5 0.81553 E 2 17.7 11.5 5.0 0.64972 E 2 18.8 12.1 3.9 0.64362 F 2 18.7 14.8 3.7 0.79144 F 2 14.7 9.9 3.6 0.67347 F 2 25.2 19.3 6.1 0.76587 D 2 23.0 17.0 5.5 0.73913 D 2 16.5 13.9 5.3 0.84242 D 2 20.8 16.5 4.8 0.79327 D 2 18.2 11.7 6.2 0.64286 D 2 16.0 10.2 4.1 0.63750 D 2 23.4 17.3 4.9 0.73932 D 2 14.8 8.8 3.9 0.59459 C 2 16.8 11.4 3.2 0.67857 C 2 14.4 10.2 3.1 0.70833 C 2 17.8 10.2 3.6 0.57303 C 2 10.9 8.8 2.4 0.80734 B 2 15.7 13.6 3.8 0.86624 B 2 15.0 11.8 3.5 0.78667 B 2 9.8 7.1 2.7 0.72449 B 2 17.9 12.7 4.7 0.70950 A 2 17.3 12.4 3.9 0.71676 A 2 21.7 17.8 4.2 0.82028 A 2 * 15.6 5.2 * A 2 18.1 12.1 4.5 0.66851 A 2 16.4 11.7 4.9 0.71341 I 2 14.9 10.8 4.5 0.72483 I 2 23.3 17.0 6.3 0.72961 I 2 18.6 14.6 5.7 0.78495 I 2 21.1 15.6 6.1 0.73934 I 2 15.0 10.2 5.9 0.68000 I 2
length width stem lobe ratio cluster tree TREE.NO 11.6 8.5 3.8 0.73276 I 2 17.4 11.1 5.7 0.63793 H 2 14.1 9.3 4.3 0.65957 H 2 19.7 14.4 5.4 0.73096 H 2 16.3 11.2 3.9 0.68712 H 2 19.6 15.4 5.0 0.78571 H 2 13.4 10.1 3.8 0.75373 H 2 20.7 16.7 5.3 0.80676 H 2
Box plots of lengths by cluster number (for tree #2)
Box plots of ratios by cluster number (for tree #2)
Return to STATS Project Reports Index