Dynamics of Green Ash Woodlands
in Theodore Roosevelt National Park
Results
Overall changes
In 1985, the tree canopies in all of the sampled stands were dominated by green ash with choke cherry as the dominant understory tree. Saskatoon service-berry was the third most abundant species in the tree stratum. This pattern had not changed by 1996. Plots of mean stem counts for the tree stratum by year (Fig. 4) indicated little change occurred in green ash, but choke cherry declined through the sample period, and service-berry declined sharply between 1988 and 1992. Between 1985 and 1996, median stem density declined 53% for choke cherry and 71% for Saskatoon service-berry (Table 3).Figure 4. Mean stems per m² and standard errors (horizontal lines above bars) for four plant categories in the tree stratum recorded in 12 green ash stands in the South Unit of Theodore Roosevelt National Park, 1985 to 1996. |
Table 3. Wilcoxin's tests of changes in coverage (ground stratum categories) and density (shrub and tree strata) between 1985 and 1996 in 12 green ash stands monitored in Theodore Roosevelt National Park. Species or genera included in limited multi-species categories are identified in Table 1. | ||||
Category | Median | Summed
ranks 1985 positive |
P | |
1985 | 1996 | |||
Tree density (N/60 m², > 200 cm) | ||||
All trees | 64.5 | 56.5 | 62.0 | 0.08 |
Green ash | 15.5 | 13.5 | 32.0 | 0.96 |
Choke cherry | 21.5 | 10.0 | 64.0 | 0.05 |
Saskatoon service-berry | 3.5 | 1.0 | 40.0 | 0.04 |
Shrub density (N/7 m², 0-200 cm) | ||||
Green ash (0 - 200 cm) | 4.5 | 3.0 | 45.0 | 0.08 |
Choke cherry (0 - 200 cm) | 56.0 | 49.0 | 47.5 | 0.53 |
Saskatoon service-berry (0 - 200 cm) | 4.0 | 2.5 | 24.5 | 0.86 |
Snowberry (0 - 200 cm) | 27.5 | 39.5 | 34.0 | 0.97 |
All species 0 - 50 cm | 80.0 | 75.0 | 41.0 | 0.50 |
All species > 50 - 100 cm | 19.5 | 17.5 | 59.5 | 0.12 |
All species > 100 - 200 cm | 18.0 | 12.0 | 53.5 | 0.27 |
Ground stratum canopy/ground cover (%) | ||||
Bare ground | 1.0 | 0.1 | 27.0 | 0.64 |
Litter | 95.2 | 65.3 | 78.0 | < 0.01 |
Total shrubs | 5.0 | 9.4 | 33.0 | 0.67 |
Total graminoids | 3.3 | 11.7 | 6.0 | 0.01 |
Climax graminoids | 2.8 | 11.3 | 6.0 | 0.01 |
Exotic grasses | < 0.1 | 0.5 | 14.0 | 0.34 |
Total forbs | 2.1 | 6.8 | 8.0 | 0.02 |
Palatable forbs | 1.4 | 3.1 | 22.0 | 0.20 |
Invasive forbs | 0.0 | 0.3 | 0.0 | < 0.01 |
Plots of mean stem counts in the shrub stratum indicated increases through 1988 and declines or stability thereafter (Fig. 5). Median stem counts in 1985 and 1996 were not significantly different for any shrub species or height classes that we tested (Table 3).
Figure 5. Mean stems per m² and standard errors (horizontal lines above bars) for five plant categories in the shrub stratum recorded in 12 green ash stands in the South Unit of Theodore Roosevelt National Park, 1985 to 1996. |
Shrub canopy coverage in the ground stratum was highest in 1994 and lowest in 1992 and 1996 (Fig. 6). The relatively large changes between some consecutive sampling periods were due to high annual variability in choke cherry seedling numbers and/or annual differences in leaf volume on snowberry (Symphorcarpos albus L./occidentalis L.). If we had sampled only in 1985 and 1996, we would have seen no significant differences in median canopy coverage for shrubs less than 50 cm in height (Table 3).
Figure 6. Mean percent cover and standard errors (horizontal lines above bars) for eight categories in the ground stratum recorded in 12 green ash stands in the South Unit of Theodore Roosevelt National Park, 1985 to 1996. |
The most striking change in canopy coverage of herbaceous plants was the increase in graminoids between 1985 and 1996 (Fig. 6). The increase was, however, not constant. A decline occurred in 1990, but coverage increased again in 1992. Climax graminoids (Table 1) and exotic grasses increased over the 12-year period, but the ratio of climax to exotic species remained highly skewed in favor of native climax species. Bare ground and forb cover remained relatively low throughout the sampling period. Palatable forbs did not increase as much as invasive forbs, but invasive forb cover remained low in 1996. When medians for 1985 and 1996 were compared (Table 3), total graminoids, climax graminoids, total forbs, and invasive forbs showed significant increases.
Multiple t-tests of truncated percent change between 1985 and 1996 for 12 stands with zero to 12 years of access by elk (Table 4) indicated no overall difference among stands (Hotelling's t² = 36.63, P = 0.35) for the 18 vegetation/ground cover variables we included in the test. There were differences (P < 0.05) in 13 of 66 tests of stand pairs but no consistent pattern supporting a relationship between longer elk use of a site and greater declines in vegetation indices between 1985 and 1996. The mean percent change over all variables for Site 6, the only stand with no elk use between 1985 and 1996, was +14%. This mean was significantly higher (P < 0.05) than means for two sites with 10 or more years of elk use (Site 9 = 20% and Site 13 = 9%), but it was not significantly different from means for 6 other sites with 10 or more years of elk use (range of means = 15% to +26%).
Changes between individual sampling periods
Differences in density and canopy cover between individual sampling periods (Figs. 4 to 6) indicated that vegetation might be responding to short-term rather than long-term factors. We developed a matrix of 224 Pearson correlations (14 weather, phenology, and short-term ungulate use indices by 16 vegetation variables). The highest correlation coefficient ( r ) in the matrix was 0.55 indicating that no single weather or ungulate index provided a good explanation of changes in individual vegetation variables in the stands (Table 5). When we looked at the patterns of moderate (P = 0.05 to 0.10) and strong (P < 0.05) associations across all vegetation variables, one phenology index and two weather indices were significantly correlated with changes in half or more of the vegetation variables: 1) the month in which vegetation measurements were made (10 significant negative correlation coefficients indicating greater negative changes for late summer vegetation measurements than for early summer vegetation measurements); 2) precipitation in the growing season the year before vegetation measurements were made (nine significant positive associations indicating greater positive changes with greater precipitation); and 3) the sum of growing season precipitation in the year prior to and the year of vegetation measurement (eight of nine significant correlation coefficients suggested positive changes with higher precipitation). Only one significant association was indicated between vegetation variables and elk pellet cover and two between vegetation and deer pellet cover.Table 5. Correlations between independent variables and percent change (truncated to -100% to +101%) in 16 vegetation variables measured at two-year intervals in Theodore Roosevelt National Park, 1986 to 1996. | ||||||||||||||
Month | Ppt- mo |
Temp- mo |
Ppt- gro |
Ppt- prev |
Ppt- 2y |
Snow | Snow- 2y |
Tem- sum |
Tem- sum2y |
Win- tem |
Win- tem2y |
Elk- pel |
Deer- pel |
|
Shrub stratum (stem counts) | ||||||||||||||
Green ash | 0.00 | -0.21a | 0.17 | 0.08 | 0.19 | 0.22a | 0.01 | 0.15 | 0.06 | -0.16 | -0.14 | -0.25a | 0.06 | -0.17 |
Choke cherry | -0.36b | 0.07 | 0.00 | -0.25a | 0.49b | 0.29a | -0.05 | 0.05 | 0.11 | -0.14 | 0.08 | 0.15 | -0.08 | 0.25b |
Saskatoon service-berry | -0.02 | -0.20 | 0.25a | 0.09 | 0.04 | 0.10 | 0.15 | 0.22a | 0.13 | 0.00 | -0.23a | -0.27b | 0.09 | 0.01 |
Snowberry | -0.23a | -0.23a | 0.26b | -0.05 | 0.55b | 0.06 | 0.25a | 0.17 | -0.26b | -0.18 | -0.2a | 0.02 | -0.15 | |
Tot 0-50 |
-0.32b | -0.11 | 0.17 | -0.17 | 0.54b | 0.05 | 0.20 | 0.19 | -0.18 | -0.07 | -0.06 | -0.13 | -0.09 | |
Tot 51-100 |
-0.13 | -0.02 | -0.06 | 0.04 | 0.08 | 0.10 | -0.10 | -0.10 | -0.13 | -0.21 | 0.05 | -0.07 | 0.13 | -0.02 |
Tot 101-200 |
-0.11 | 0.13 | 0.04 | 0.09 | 0.08 | 0.13 | 0.10 | -0.09 | -0.13 | -0.13 | -0.08 | -0.02 | 0.04 | 0.04 |
Ground stratum (% cover) | ||||||||||||||
Bare ground | -0.27b | 0.32b | -0.07 | -0.49b | 0.10 | -0.23a | 0.25a | 0.10 | 0.34b | 0.51 | 0.12 | 0.56b | -0.10 | 0.34b |
Litter | 0.16 | 0.08 | 0.34b | 0.23a | -0.11 | 0.05 | 0.18 | 0.21 | 0.11 | 0.16 | -0.39b | -0.21a | -0.03 | -0.08 |
Low shrub canopy (0-50 cm) |
-0.50b | -0.11 | 0.06 | -0.26a | 0.55b | 0.00 | 0.10 | 0.12 | -0.25a | 0.04 | 0.00 | 0.02 | -0.03 | |
Total graminoids | -0.43b | 0.24a | -0.04 | -0.45b | 0.55b | 0.21 | 0.04 | 0.06 | 0.23a | 0.07 | 0.13 | 0.42b | -0.24a | 0.21 |
Climax graminoids | -0.38b | 0.16 | -0.07 | -0.33b | 0.42b | 0.18 | -0.02 | 0.00 | 0.12 | -0.02 | 0.14 | 0.29b | -0.06 | 0.17 |
Exotic grasses | -0.36b | 0.04 | 0.14 | -0.27b | 0.42b | 0.22a | 0.06 | 0.18 | 0.23a | 0.02 | -0.07 | 0.10 | -0.04 | 0.00 |
Total forbs | -0.36b | 0.04 | 0.26b | -0.04 | 0.33b | 0.12 | 0.21 | 0.16 | -0.07 | -0.22a | -0.13 | 0.08 | -0.14 | |
Palatable forbs | -0.24a | 0.05 | 0.41b | -0.04 | 0.50b | 0.08 | 0.33b | 0.27b | -0.04 | -0.35b | -0.15 | 0.02 | -0.12 | |
Invasive forbs | 0.00 | 0.05 | 0.17 | 0.21 | -0.16 | -0.01 | 0.14 | 0.05 | -0.04 | 0.02 | -0.23a | -0.20 | 0.12 | -0.05 |
a p
< 0.10 b p < 0.05 |
In our multivariate analysis, we used the same vegetation categories and independent variables used to develop the Pearson correlation matrix, but we created a separate logistic regression model for each vegetation variable (Table 6). Only five of the 14 independent variables (weather and ungulate indices) met the minimum criterion for entry into any model (P < 0.05). No models included more than two independent variables. The SAS (1994) Logistic Regression module created statistically significant models for predicting positive or negative changes in 12 of the 16 vegetation categories. Of the 12 models in which inclusion of independent variables was statistically justifiable, only eight could be considered biologically significant, i.e., the contribution of the selected model to prediction of positive or negative changes was greater than 25% more than that of the null, no variable, model. In seven of these eight models (models for prediction of change in choke cherry stems, snowberry stems, shrub stems greater than 100 to 200 cm, low shrub canopy cover, graminoid canopy cover, exotic grass canopy cover, and palatable forb canopy cover) precipitation in the growing season of the year prior to vegetation measurement, one of the variables with the highest number of significant associations in the univariate analysis, was the only variable in the model or the first variable to enter. Month of measurement and summed precipitation in the growing season the year before and the year of vegetation measurement, two other important variables identified in the univariate analysis, did not enter in any model.
Table 6. Summary of stepwise logistic regression analysis of changes in vegetation categories in the shrub and ground strata for measurements in Theodore Roosevelt National Park at 2-year intervals from 1986 through 1996. A positive value for the standardized coefficient of individual variables to the logit indicates that an increase in that independent variable would predict an increase in the associated vegetation category. A negative value would predict an inverse relationship. No model included more than two of the set of 16 independent variables in the analysis. | ||||||||||
Category | Changes (n) | Significant (P < 0.05) Variables | AICa | Model R2 |
Hosmeyer- Lemeshow test of P-value |
|||||
+ (or 0) |
| First variable to enter |
Standardized coefficient | Second variable to enter |
Standardized coefficient | Intercept only | Full model | |||
Shrub stratum (stem counts) | ||||||||||
Green ash | 33 | 27 | None | |||||||
Choke cherry | 25 | 35 | Pptprev | 0.666 | 85.5 | 69.3 | 0.32 | 0.75 | ||
Saskatoon Service-berry | 34 | 26 | None | |||||||
Snowberry | 25 | 35 | Pptprev | 0.785 | Temwin2y | -0.415 | 83.5 | 63.4 | 0.45 | 0.96 |
All species 0-50cm | 30 | 30 | None | |||||||
All species 50-100cm | 26 | 34 | Temsum | -0.385 | 84.1 | 79.9 | 0.13 | 0.61 | ||
All species 100-200cm | 21 | 39 | Pptprev | 0.600 | 79.7 | 69.0 | 0.26 | 0.53 | ||
Ground stratum (% cover) | ||||||||||
Bare ground | 32 | 28 | Pptmo | 0.714 | Temsum2y | 0.927 | 84.9 | 63.2 | 0.46 | 0.09 |
Litter | 17 | 43 | Temwin2y | -0.494 | 73.5 | 68.0 | 0.17 | 0.42 | ||
Shrubs 0-50cm | 24 | 36 | Pptprev | 0.758 | 82.8 | 65.3 | 0.37 | 0.62 | ||
Graminoids | 39 | 21 | Pptprev | 0.525 | Temwin2y | 0.409 | 79.7 | 68.2 | 0.31 | 0.69 |
Climax graminoids | 35 | 25 | Pptprev | 0.422 | Temwin2y | 0.345 | 83.5 | 75.6 | 0.24 | 1.00 |
Exotic grasses | 43 | 17 | Pptprev | 0.654 | 73.5 | 62.6 | 0.28 | 0.58 | ||
All forbs | 27 | 33 | Snow | 0.360 | 84.6 | 81.2 | 0.10 | 0.67 | ||
Palatable forbs | 28 | 32 | Pptprev | 0.584 | 84.9 | 73.4 | 0.27 | 0.43 | ||
Invasive forbs | 53 | 7 | None | |||||||
a AIC = Akaike Information Content is a relative measure of the efficiency of the model. The lower the value, the more efficient the model. If the AIC for the null model (intercept only) is ≥2 units more than the model with independent variables added, the full model is significantly (P < 0.05) better than the null model. |
Previous Section -- Methods
Return to Contents
Next Section -- Discussion