Results

1 Survival of Planted Trees and Shrubs

Planted native shrub and tree survival was fairly low. Median survival proportions ranged from 0.20 (Fig. 1.1). Woody plant survival depended on whether weed fabric was used during revegetation, the growth form of the individual, and revegetation treatment. There was not a significant interaction between weed fabric and revegetation treatment (\(\chi^2=0.11\), \(p=0.95\)) so the interaction term was removed (Table 1.1). Holding growth form constant, the odds of survival for individuals planted using weed fabric are 37% higher than those without. Holding the weed fabric variable constant, the odds of survival are %187 higher for trees than shrubs(Table 1.1). There was no difference in survival between shrub only and tree only revegetation treatments (Tukey’s HSD, \(z=1.77\), \(p=0.18\)) or shrub only and trees and shrubs (Tukey’s HSD, \(z=-1.93\), \(p=0.13\)), but a higher proportion of individuals planted in the trees and shrubs revegetation subplots survived than in the trees only subplots (Tukey’s HSD, \(z=-2.91\), \(p=0.01\)).

Table 1.1: Output for negative binomial model of woody plant survival by treatment, weed fabric use, and growth form.

	Likelihood of survival
Predictors	Odds Ratios	Conf. Int (95%)	p-Value	df
(Intercept)	0.32	0.21 – 0.49	<0.001	Inf
weedfabricY	1.37	1.02 – 1.85	0.036	Inf
Growth Form (Tree)	2.87	1.47 – 5.77	0.002	Inf
Restoration Treatment (Trees Only)	0.48	0.21 – 1.07	0.076	Inf
Restoration Treatment (Trees and Shrubs)	1.42	0.99 – 2.02	0.053	Inf
Plot	1.09	0.96 – 1.24	0.181	Inf
Observations	783
R2 Tjur	0.033

Figure 1.1: Proportion of planted trees and shrubs that survived until 2021 separated by restoration treatment. Those planted using weed fabric are in red and no weed fabric are tan.

2 Functional Group Cover

Grasses showed variability in response to replanting treatment over time. There was a significant interaction between replanting treatment and year of sampling for annual bromes (\(\chi^2=248.79\), \(p<0.001\);). There was large variability in Brome cover among treatments over time with some replanting treatments higher than the unplanted control in some years and lower in others. Differences among replanting treatments we also highly variable over time (Figure 2.1). There was no significant interaction between treatment and year for invasive perennial grasses so we focused on main effects. Restoration treatment was a significant predictor (\(\chi^2=112.44\), \(p<0.001\)). All replanting treatments had lower invasive pernnial grass cover than the unplanted controls (control-herbaceous \(z=5.27\), \(p<0.001\); control-herb and shrub \(z=4.37\), \(p=0.001\); control-herb and tree \(z=2.77\), \(p=0.04\), control-herb, shrub, and tree \(z=8.6\),\(p<0.001\)) The herbaceous, shrub, and tree replanting treatment had lower invasive perennial grass cover than all of the other replanting treatments (Figure 2.1).

Figure 2.1: Differences between replanting treatments and unplanted controls in grass cover over the course of the study broken down into functional group (annual brome, left column; invasive perennial grasses, center; native perennial grasses, right column)

Forb cover exhibited less overall variability in response to replanting treatment over time than grass cover. There was not a significant interaction between replanting treatment and year of sampling for annual bromes (\(\chi^2=248.79\), \(p<0.001\);). There was large variability in Brome cover among treatments over time with some replanting treatments higher than the unplanted control in some years and lower in others. Differences among replanting treatments we also highly variable over time (Figure 2.2). There was no significant interaction between treatment and year for invasive perennial grasses so we focused on main effects. Restoration treatment was a significant predictor (\(\chi^2=112.44\), \(p<0.001\)). All replanting treatments had lower invasive pernnial grass cover than the unplanted controls (control-herbaceous \(z=5.27\), \(p<0.001\); control-herb and shrub \(z=4.37\), \(p=0.001\); control-herb and tree \(z=2.77\), \(p=0.04\), control-herb, shrub, and tree \(z=8.6\),\(p<0.001\)) The herbaceous, shrub, and tree replanting treatment had lower invasive perennial grass cover than all of the other replanting treatments (Figure 2.2).

Figure 2.2: Differences between replanting treatments and unplanted controls in forb cover over the course of the study broken down into functional group (invasive forbs, left column; native forbs, right column)

Woody plant cover also exhibited less overall variability in response to replanting treatment over time than grass cover, although there was some variability in tree cover over time. There was not a significant interaction between replanting treatment and year of sampling for native shrubs but there was a significant treatment effect (\(\chi^2=45.4\), \(p<0.001\)). The herbaceous only planting had lower native shrub cover than unplanted controls (\(z=3.15\),\(p=0.014\)) and the tree and herbaceous planting treatment also had less shrub cover than controls(\(z=2.86\), \(p=0.035\)).Multiple comparisons showed that treatments that included shrubs in the planting had higher shrub cover than herbaceous only plantings or herbaceous and tree plantings(Figure 2.3). Year was also a significant variable in the model (\(\chi^2=20.16\), \(p=0.014\)) with decreasing shrub cover in the last several years of sampling. There was no significant interaction between treatment and year for native tree cover, but both main effects were significant (treatment \(\chi^2=37.5\), \(p<0.001\); year \(\chi^2=9.6\), \(p<0.001\)). Herbaceous only planting treatment had lower tree cover than unplanted controls (\(z=-2.92\), \(p=0.004\)) while herbaceous, shrub, and tree plantings had greater tree cover than unplanted controls (\(z=3.41\), \(p<0.001\)) and all other planting treatments according to multiple comparisons (Figure 2.3). 2014 (\(z=-3.26\), \(p=0.004\)) and 2015 (\(z=-3.24\), \(p<0.001\)) had lower native tree cover than pretreatment sampling in 2010 and 2012 had lower native tree cover than most other years. There was also no significant interaction for plains cottonwood cover but significant treatment (\(\chi^2=32.34\), \(p<0.001\)) and year (\(\chi^2=26.43\), \(p=0.003\)) main effects. Only the herbaceous, shrub, and tree treatment had higher plains cottonwood cover than the controls (\(z=-4.02\), \(p=0.001\)). Mutltiple comparisons showed that it was also higher than other planting treatments. 2012 was the only year with significantly lower cottonwood cover than several other years 2.3).

Figure 2.3: Differences between replanting treatments and unplanted controls in woody plant cover over the course of the study broken down into functional group (native shrubs, left column; native trees, center; plans cottonwood, right column)

3 Seedling emergence

There was high year-to-year variability in the number of seedlings that germinated in the plots. There was no interaction between year and replanting treatment and there were no differences among replanting treatments (\(\chi^2=27.2\), \(p=0.71\)). Year was a significant factor in predicting the number of seedlings (\(\chi^2=97.89\), \(p<0.001\)). Tukey’s analysis showed 2016 and 2020 to have higher seedling counts than most others (Figure 3.1).

Figure 3.1: Total number of seedlings that emerged and were subsequently removed via hand weeding or herbicide in each year of the study for each restoration and planting treatment