Historical Disturbance Trends: Round 3 Results

3rd time is the charm?
Author

Madeleine Desrochers

Published

June 21, 2023

What’s Changed Since Last Time?

In order to assess what is going on in 2021, we have added a year to the study period. These results now cover from 1990 - 2022. The high values seen in all metrics for 2021 do not appear to be the result of it being at the end of the previous time period as we do not see similarly high values for 2022. Additionally, for each metric that showed a trend with MK tests now also has results from an ANOVA test.

Included in this post:

  • Results for the entire ADK FP 1990 - 2022

  • Results for Age Group Divisions

  • Age Group ANOVA

Forest Preserve Wide Man Kendall Test Results for Patch Metrics

Metric Man Kendall Statistic Normalized Test Statistic p Value Trend Sens Slope
area -0.1212 -0.9761 0.3290 no trend -0.0020
cai 0.0720 0.5733 0.5664 no trend 0.0027
circle 0.0720 0.5733 0.5664 no trend 0.0027
contig 0.0720 0.5733 0.5664 no trend 0.0027
core -0.0720 -0.5733 0.5664 no trend -0.0001
enn 0.4432 3.6102 0.0003 increasing 2.5198
frac -0.3258 -2.6495 0.0081 decreasing -0.0004
gyrate -0.1439 -1.1621 0.2452 no trend -0.0644
ncore -0.1212 -0.9761 0.3290 no trend -0.0006
para -0.3030 -2.4636 0.0138 decreasing -0.0001
perim -0.1705 -1.3790 0.1679 no trend -1.6070
shape -0.2917 -2.3706 0.0178 decreasing -0.0033
mean_mag 0.3144 2.5566 0.0106 increasing 0.8322
percent_area_dist -0.3295 -2.6805 0.0074 decreasing -0.0083

Metric Plots

All patch metrics derived from the landscape metrics package in r. Descriptions of the patch metrics can be found here.

Average disturbance patch size annually from 1990 - 2022.

Average core area index for disturbance patches from 1990 - 2021

Average size of smallest circumscribing circle for disturbance patches 1990 -2021

Average contiguity index for disturbance patches 1990 - 2021

Average size of patch core area from 1990 - 2021

Average distance to closest neighboring patch annually from 1990 - 2021

Average fractal dimension index from 1990 - 2021

Average radius of gyration for disturbance patches from 1990 - 2021

Average number of core areas in one disturbance patch from 1990 - 2021

Average perimeter to area ratio for disturbance patches 1990 - 2021

Average length of disturbance patch perimeter from 1990 - 2021

Average shape index for disturbance patches 1990 - 2021

Average disturbance magnitude of disturbance 1990 - 2021

Percentage of forest preserve land disturbed annually from 1990 - 2021

Man Kendall Test Results Split by Age Group

Forest preserve properties purchased prior to 1990 were split into 4 groups based on the time since acquisition by the state of New York. Included in this analysis are Forest Preserve parcels with known dates of acquisition and parcels documented on state land survey maps from the APA for 1893 and 1919. Additionally, only parcels acquired before 1990 were used so that there were no parcels acquired mid analysis. In total, 1364 parcels representing 697,818 ha of forest preserve land were included. Age groups were created by assigning roughly equal areas of forest preserve to each bin (~174,000 ha per group).

Group 1: Aquistion prior to 1893

Metric Man Kendall Statistic Normalized Test Statistic p Value Trend Sens Slope
area -80 -1.2241 0.2209 no trend -0.0029
cai 46 0.6972 0.4856 no trend 0.0044
circle 46 0.6972 0.4856 no trend 0.0044
contig 46 0.6972 0.4856 no trend 0.0044
core -54 -0.8212 0.4115 no trend -0.0002
enn 254 3.9201 0.0001 increasing 3.0322
frac -174 -2.6805 0.0074 decreasing -0.0004
gyrate -84 -1.2860 0.1984 no trend -0.0924
ncore -60 -0.9142 0.3606 no trend -0.0007
para -172 -2.6495 0.0081 decreasing -0.0001
perim -112 -1.7199 0.0855 no trend -2.7480
shape -154 -2.3706 0.0178 decreasing -0.0034
mean_mag 160 2.4636 0.0138 increasing 0.8560
percent_area_dist -218 -3.3623 0.0008 decreasing -0.0104

Group 2: Acquisition between 1894 and 1918

Metric Man Kendall Statistic Normalized Test Statistic p Value Trend Sens Slope
area -50 -0.7592 0.4477 no trend -0.0015
cai 54 0.8212 0.4115 no trend 0.0054
circle 54 0.8212 0.4115 no trend 0.0054
contig 54 0.8212 0.4115 no trend 0.0054
core -2 -0.0155 0.9876 no trend 0.0000
enn 212 3.2693 0.0011 increasing 2.6804
frac -138 -2.1227 0.0338 decreasing -0.0004
gyrate -80 -1.2241 0.2209 no trend -0.0675
ncore -4 -0.0465 0.9629 no trend 0.0000
para -96 -1.4720 0.1410 no trend -0.0001
perim -94 -1.4410 0.1496 no trend -1.7930
shape -106 -1.6269 0.1038 no trend -0.0029
mean_mag 156 2.4016 0.0163 increasing 0.8717
percent_area_dist -170 -2.6186 0.0088 decreasing -0.0068

Group 3: Acquisition between 1919 and 1956

Metric Man Kendall Statistic Normalized Test Statistic p Value Trend Sens Slope
area -62 -0.9452 0.3446 no trend -0.0019
cai 0 0.0000 1.0000 no trend 0.0000
circle 0 0.0000 1.0000 no trend 0.0000
contig 0 0.0000 1.0000 no trend 0.0000
core -4 -0.0465 0.9629 no trend 0.0000
enn 218 3.3623 0.0008 increasing 2.5758
frac -158 -2.4326 0.0150 decreasing -0.0003
gyrate -56 -0.8522 0.3941 no trend -0.0530
ncore -42 -0.6353 0.5253 no trend -0.0006
para -88 -1.3480 0.1777 no trend -0.0001
perim -92 -1.4100 0.1585 no trend -1.7083
shape -154 -2.3706 0.0178 decreasing -0.0024
mean_mag 124 1.9058 0.0567 no trend 0.7234
percent_area_dist -154 -2.3706 0.0178 decreasing -0.0066

Group 4: Acquisition between 1957 and 1989

Metric Man Kendall Statistic Normalized Test Statistic p Value Trend Sens Slope
area -38 -0.5733 0.5664 no trend -0.0010
cai 92 1.4100 0.1585 no trend 0.0099
circle 92 1.4100 0.1585 no trend 0.0099
contig 92 1.4100 0.1585 no trend 0.0099
core -6 -0.0775 0.9382 no trend 0.0000
enn 236 3.6412 0.0003 increasing 2.4103
frac -120 -1.8438 0.0652 no trend -0.0003
gyrate -56 -0.8522 0.3941 no trend -0.0384
ncore 16 0.2324 0.8162 no trend 0.0004
para -140 -2.1537 0.0313 decreasing -0.0001
perim -74 -1.1311 0.2580 no trend -1.2197
shape -116 -1.7819 0.0748 no trend -0.0021
mean_mag 180 2.7735 0.0055 increasing 0.8864
percent_area_dist -168 -2.5876 0.0097 decreasing -0.0061

Metric Plots

Average Disturbance Area by Acquisition Age Group. No trends in any group.

Average Distance to the Closest Disturbance by Acquisition Age Group. Increasing trends in all groups.

Average Fractal Dimension Index by Acquisition Age Group. Decreasing trends for groups 1, 2 and 4. No trend in group 3.

Average Shape Index by Acquisition Age Group. Decreasing trends for groups 1 and 3. No trend in groups 2 and 4.

Average Perimeter to Area Ratio by Acquisition Age Group. Decreasing trends for groups 1 and 4. No trends in groups 2 and 3.

Average Disturbance Magnitude by Acquisition Age Group. Increasing trends for groups 1, 2 and 4. No trend in group 3.

Percentage of Forest Preserve Land Disturbed Annually by Acquisition Age Group. Decreasing trends for all groups.

Age Group ANOVA

ENN

Analysis of Variance Table

Response: enn
              Df     Sum Sq Mean Sq F value    Pr(>F)    
bin            3     659339  219780  13.036 1.651e-08 ***
Residuals 116212 1959205350   16859                      
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = enn ~ bin, data = grouped_patches)

$bin
           diff       lwr      upr     p adj
2-1 -0.05050715 -2.798371 2.697357 0.9999622
3-1  0.71483404 -1.777375 3.207043 0.8822446
4-1  6.53703195  3.676821 9.397243 0.0000000
3-2  0.76534119 -2.210126 3.740808 0.9117630
4-2  6.58753910  3.297690 9.877388 0.0000016
4-3  5.82219791  2.742676 8.901720 0.0000071

For the ENN metric, these tests suggest that group 4 (1957 - 1989) is significantly different from all other groups, but the other groups are not significantly different from each other.

Fractal Dimension Index

Analysis of Variance Table

Response: frac
              Df Sum Sq   Mean Sq F value Pr(>F)
bin            3   0.01 0.0036370  1.2845 0.2777
Residuals 116212 329.04 0.0028314               
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = frac ~ bin, data = grouped_patches)

$bin
             diff           lwr          upr     p adj
2-1  0.0001128933 -0.0010132156 0.0012390023 0.9940145
3-1  0.0002278391 -0.0007934992 0.0012491774 0.9401303
4-1 -0.0006898416 -0.0018619917 0.0004823085 0.4302421
3-2  0.0001149458 -0.0011044377 0.0013343292 0.9950095
4-2 -0.0008027349 -0.0021509559 0.0005454861 0.4196602
4-3 -0.0009176807 -0.0021797070 0.0003443457 0.2418770

For the Fractal Dimension Index, these tests suggest that there is no difference between the four age groups.

Perimeter to Area ratio

Analysis of Variance Table

Response: para
              Df Sum Sq   Mean Sq F value   Pr(>F)    
bin            3  0.016 0.0054958  15.854 2.65e-10 ***
Residuals 116212 40.284 0.0003466                     
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = para ~ bin, data = grouped_patches)

$bin
             diff           lwr           upr     p adj
2-1 -0.0005666264 -9.606495e-04 -1.726033e-04 0.0012609
3-1 -0.0009402585 -1.297623e-03 -5.828944e-04 0.0000000
4-1 -0.0004350295 -8.451623e-04 -2.489669e-05 0.0326008
3-2 -0.0003736321 -8.002917e-04  5.302758e-05 0.1100960
4-2  0.0001315970 -3.401427e-04  6.033366e-04 0.8905214
4-3  0.0005052290  6.364872e-05  9.468093e-04 0.0173283

For the Perimeter to Area ratio metric, these tests suggest that group 3 (1919 - 1956) is significantly different from all other groups, but the other groups are not significantly different from each other.

Shape

Analysis of Variance Table

Response: shape
              Df Sum Sq Mean Sq F value   Pr(>F)   
bin            3      4 1.31650  3.8112 0.009601 **
Residuals 116212  40143 0.34543                    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = shape ~ bin, data = grouped_patches)

$bin
            diff          lwr         upr     p adj
2-1  0.005749251 -0.006689051 0.018187554 0.6347910
3-1  0.014630164  0.003349094 0.025911234 0.0047871
4-1  0.003106984 -0.009839860 0.016053828 0.9268550
3-2  0.008880913 -0.004587642 0.022349467 0.3268006
4-2 -0.002642267 -0.017533882 0.012249347 0.9685108
4-3 -0.011523180 -0.025462741 0.002416381 0.1455298

For the Shape metric, these tests suggest that groups 1 (pre 1893) and 3 (1919 - 1956) are significantly different from each other, but not significantly different from the other groups.

Mean Magnitude

Analysis of Variance Table

Response: mean_mag
              Df    Sum Sq Mean Sq F value    Pr(>F)    
bin            3    888340  296113  131.97 < 2.2e-16 ***
Residuals 116212 260763322    2244                      
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = mean_mag ~ bin, data = grouped_patches)

$bin
        diff       lwr      upr     p adj
2-1 2.078648 1.0761612 3.081134 0.0000006
3-1 3.513163 2.6039452 4.422380 0.0000000
4-1 7.926428 6.8829549 8.969902 0.0000000
3-2 1.434515 0.3489934 2.520036 0.0038312
4-2 5.847781 4.6475651 7.047996 0.0000000
4-3 4.413266 3.2897826 5.536749 0.0000000

For mean patch magnitude, these tests suggest that all groups are significantly different from all other groups.