Plant community response to fuel break and goat grazing in southern California

31 Background: California is a global biodiversity hotspot, yet increased urbanization of wildlands, warming 32 temperatures, and invasion of nonnative species pose serious risks to these areas due to an increase in 33 wildfire frequency. Fuel management is a tool for reducing fire risk to neighboring communities and 34 natural resources that involves a two-step process requiring an initial reduction of woody vegetation 35 followed by a repeated control of woody plants and reduction of herbaceous cover. To understand the 36 compositional and structural changes resulting from fuel treatment methods in southern California 37 chaparral, we evaluated the compositional and structural impacts of a recently created fuel break 38 established around the Lake Morena community on the Cleveland National Forest. The area was initially 39 treated with cut and pile burning, then treated with herbicide, and lastly grazed by 1,200 goats. The 40 purpose of this study is to (1) evaluate the compositional and structural differences associated with the 41 initial fuel break, and (2) quantify compositional shifts in herbaceous and woody vegetation caused by 42 goat grazing over time. 43 Results: Plots on fuel breaks and in adjacent wildlands exhibited significantly different species 44 assemblages. Total herbaceous cover (both native and nonnative) was 92 times greater on fuel breaks 45 than in adjacent chaparral-dominated wildlands and native shrub cover was 55.3 times greater in 46 adjacent wildlands than on fuel breaks. Goats had a significant impact on reducing native and nonnative 47 herb cover (87% reduction in cover, 92% reduction in height), but were ineffective at reducing the cover 48 and height of most woody species such as Adenostoma fasciculatum, Eriogonum fasciculatum, Quercus 49 berberidifolia, and Artemisia tridentata. However, goats were found to be effective in controlling 50 nonnative grasses including Bromus diandrus and Bromus madritensis. 51 Conclusion: Initial fuel break creation was effective at reducing wood biomass and height, 52 simultaneously giving rise to an abundance and diversity of native and nonnative herbaceous species.

30-meter transect line and averaged for untreated, pre-grazing, and post-grazing treatment types. 124 Lifeform cover and species richness were summed across the entire transect and averaged for 125 untreated, pre-grazing, and post-grazing plots. The tallest individual at each sampling point was 126 recorded as part of the point-line intercept method. Lifeform height was calculated across each transect 127 as the sum of herb or shrub height divided by the total number of transect points where that particular 128 lifeform was the tallest. 129 To characterize ground cover and fuels, a one-square meter quadrat was used at five locations 130 (5, 10, 15, 20, 25 meters) along the upslope side of the transect line. Within each quadrat, fuel height 131 was measured at five points as the distance from the soil surface to the top of the tallest unrooted dead 132 vegetation, and the ground cover (bare ground, litter, wood (above 6.35 mm diameter), live vegetation, 133 and goat feces) was visually estimated. Ground cover and fuel height were pooled from the five 134 quadrats to obtain a plot-level average. 135 Statistical Analysis: Non-metric multidimensional scaling (NMDS) was used to visualize 136 compositional differences between treatment (fuel break and control) as a part of the 'vegan' package 137 in R (Oksanen et al. 2011). The ordination uses rank-order correlation and Bray-Curtis dissimilarities, 138 with the metaMDS function, to model the differences among treatment and control plots based on 139 species composition and abundance of all plant species. Two sample t-tests were used to evaluate 140 differences in lifeform cover, richness, height, and ground cover between untreated (N=8) and treated 141 plots (N=16). Data were checked for normality using QQ plots, and the equality of variances between 142 the two groups were assessed using an F-test. Ground cover variables (bare ground, litter, and wood), 143 shrub height, and shrub and herb cover and richness were square-root transformed to meet the 144 assumptions of the t-test. Live tree richness and cover in addition to fuel and herb height were analyzed 145 using a non-parametric Wilcoxon rank-sum test because they could not be adequately transformed to 146 meet the assumptions of normality. 147 Paired t-tests were used to evaluate the effects of goat grazing on lifeform cover, richness, 148 height, and ground cover. The observed differences from pre-grazing and post-grazing treatment plots 149 were checked for normality with QQ plots. Variables were square-root (tree, wood, vegetation, native 150 herb, nonnative herb, and feces cover, and native herb, nonnative herb and dead shrub richness) or sin 151 (litter cover and live shrub richness) transformed to meet statistical assumptions. Live tree richness and 152 average herb height were analyzed using a Wilcoxon rank-sum test because they did not meet 153 assumptions of normality. All statistical analyses were done in Rstudio version 1.1.453 (Vienna,154 Austria) at α=0.05 and we report means ± 1 standard error (SE). 155

Discussion 205
This study demonstrates the complexity of fuel break creation and maintenance that involve a 206 stepwise process in vegetation change. Initial fuel break creation has significant effects on vegetation 207 cover and richness, as we expected. We found compositional differences driven by a decrease in the 208 abundance of shrubs and a higher abundance and diversity of herbaceous species inside of the fuel 209 break compared to non-disturbed areas. Chaparral shrublands are known for exhibiting unparalleled 210 temporal diversity with substantial herbaceous richness (e.g. fire followers) being expressed following 211 wildfire (Keeley et al. 2005). Herbaceous species can be triggered by various processes associated with 2001). The process of fuel reduction may mimic some of the post-fire processes by increasing 214 temperature at the soil surface or via scarification caused by ground disturbing fuels reduction activities. 215 Shrub removal may also lead to light associated cues for germination (Le Maitre and Brown 1992, Stone 216 1957). It is important to note that this study was conducted on a relatively new fuel break and the 217 difference in richness and cover for older fuel breaks are not likely to track our findings. 218 The fuel break complex at Lake Morena was recently opened in 2015 and we propose that fuel 219 break creation promotes an increase in native and nonnative species initially, but repeated disturbances 220 may lead to degradation that includes an increase nonnative annual species at the expense of native 221 species. The dominance of nonnative annuals, especially grasses, may be reinforced in frequently 222 disturbed areas through higher germination rates, competitive superiority, and accumulation of a While the Lake Morena fuel break complex is currently exhibiting signs of increased herbaceous 233 diversity, with repeated maintenance we suspect species diversity will decline and nonnative species will 234 become the dominant vegetation type in this area. Alternatively, if this fuel break is not maintained and 235 lacks future disturbance, herbaceous richness and cover are likely to decline due to the regrowth of shrubs and the exclusion of many nonnative and native annuals that were able to persist immediately 237 after disturbance. 238 Controlling woody regrowth is vital to the functioning of fuel breaks in reducing fire risk to human-239 dominated landscapes. Therefore, understanding the effectiveness of alternative maintenance methods, 240 such as goat grazing, is imperative. We found that targeted grazing was ineffective at reducing the 241 height and cover of most woody vegetation, apart from Eriophyllum confertiflorum and Cercocarpus 242 betuloides. Goats showed high selectivity when browsing shrubs and had a low preference for many of 243 the dominant shrub species at Lake Morena. This selectivity should be considered when determining if 244 goats are an economically feasible alternative to other methods of controlling regrowth. 245 Goat grazing strongly reduced both native and nonnative herbaceous cover and height, which may 246 be desirable in areas where flashy fuels and ignition risk are a concern. It is important to note that goats 247 also removed most of the nonnatives on the landscape (87% reduction) but tended to leave a higher 248 cover of Bromus tectorum L.. Of the 3.2% of the nonnative herbaceous cover remaining after grazing, habitats. Initial fuel break creation through cut and pile and herbicide application at Lake Morena was 272 effective at reducing woody biomass and height, while simultaneously giving rise to an abundance and 273 diversity of native and nonnative herbaceous species. While this may superficially appear to be a win-274 win for fuels management and ecology, it is important to note that with repeated maintenance, fuel 275 breaks are likely to become increasingly dominated by undesirable nonnative species. Goat grazing was 276 ineffective at reducing height and cover of woody vegetation but was successful in reducing both native 277 and nonnative herbaceous cover and height, which may be desirable in areas where flashy fuels and 278 ignition risk are high. However, in areas where control of woody biomass is the primary goal, land 279 managers should consider the seasonality, duration, and plant species composition when contemplating 280 goats as a tool for fuel break maintenance. 281

Availability of data and materials 283
The datasets used and analyzed during the current study are available from the corresponding author 284 upon reasonable request. 285
1.00 ± 0.6 1.00 ± 0.6 0.00 ± 0.0 Eriophyllum confertiflorum (DC.) A. Gray 1.13 ± 0.4 0.00 ± 0.0 0.13 ± 0.1 295 boxes represent 50% of the data, and each whisker represents 25% of the data. Dots represent 318 outliers. When there are no outliers, the end of the whisker designates minimum and maximum 319 values. Symbols above each category denote significant differences between treatment groups 320 (****: p < 0.0001). 321 bold horizontal lines are the medians, the boxes represent 50% of the data, and each whisker 324 represents 25% of the data. Dots represent outliers. When there are no outliers, the end of the 325 whisker designates minimum and maximum values. Symbols above each category denote 326 significant differences between treatment groups (ns: p > 0.05; ****: p < 0.0001; ***: p < 0.001; 327 **: p < 0.01). 328    Fuel break creation, through pile burning and herbicide, resulted in changes in plant lifeform cover (left) and species richness (right). Species richness is the number of species counted along a 30-meter transect line. The bold horizontal lines are the medians, the boxes represent 50% of the data, and each whisker represents 25% of the data. Dots represent outliers. When there are no outliers, the end of the whisker designates minimum and maximum values. Symbols above each category denote signi cant differences between treatment groups (ns: p > 0.05; ****: p < 0.0001; ***: p < 0.001; **: p < 0.01).

Figure 4
Fuel break creation, through pile burning and herbicide, resulted in changes to mean herbaceous height (left) and shrub height (right). The bold horizontal lines are the medians, the boxes represent 50% of the data, and each whisker represents 25% of the data. Dots represent outliers. When there are no outliers, the end of the whisker designates minimum and maximum values. Symbols above each category denote signi cant differences between treatment groups (****: p < 0.0001).

Figure 5
Goat grazing resulted in changes in plant lifeform cover (left) and species richness (right). Species richness is the number of species counted along a 30-meter transect line. The bold horizontal lines are the medians, the boxes represent 50% of the data, and each whisker represents 25% of the data. Dots represent outliers. When there are no outliers, the end of the whisker designates minimum and maximum values. Symbols above each category denote signi cant differences between treatment groups (ns: p > 0.05; ****: p < 0.0001; ***: p < 0.001; **: p < 0.01).

Figure 6
Goat grazing resulted in changes to mean herbaceous height (left) and shrub height (right). The bold horizontal lines are the medians, the boxes represent 50% of the data, and each whisker represents 25% of the data. Dots represent outliers. When there are no outliers, the end of the whisker designates minimum and maximum values. Symbols above each category denote signi cant differences between treatment groups (ns: p > 0.05; ****: p < 0.0001).

Figure 7
Goat grazing was effective at reducing the distance from ground to base of crown height on Quercus agrifolia Née. Photos taken by authors A. Grupenhoff and N. Molinari.