1.0 Executive Summary

At the Cosumnes River Preserve perennial pepperweed (Lepidium latifolium L.) impacts a variety of habitats and threatens current and future restoration activities. The consequences of this type of infestation can be costly and extremely detrimental to natural areas. As a result perennial pepperweed has been widely studied at the Preserve and beyond. Early detection and rapid response has become the mantra of weed control experts and land managers as it is cost effective and minimizes the physical and biological impacts of large scale weed infestations. In the late nineties, perennial pepperweed was identified on key properties at the Preserve, and since then it has proliferated along roadsides, in riparian forests, on floodplains, in areas surrounding wetlands, and in grasslands. The Preserve’s weed management plan identifies perennial pepperweed as a species of high-ranking threat to critical habitats because of its ability to spread and its tendency towards monospecific stands (Cosumnes River Preserve 2000). A yearly inventory of perennial pepperweed on some key properties at the Preserve (The Experimental Floodplain) from 2002 to 2004 confirmed fears that the population was increasing exponentially and that a long term, large-scale eradication plan would be necessary to control the species (Cosumnes Research Group 2005). The Perennial Pepperweed Control Project, initiated in 2005, sought to integrate multiple co-occurring studies into one adaptive management plan (see Subtask 5.2) that could be implemented by Preserve managers. To this end, the project was wide ranging in its goals and outcomes, from the creation of an online virtual herbarium for the Preserve (see Subtask 3.3) to applying herbicides (see Subtask 4.1) and modeling patch establishment (see Subtask 5.1). This final report outlines the outcomes of the project into a comprehensive report that summarizes the findings of each individual task. Each task and subtask presented its own unique challenges and broadened our understanding of perennial pepperweed’s biology, ecology, population dynamics, and eradication potential. The goals of The Perennial Pepperweed Control Project can be organized into three distinct phases:

Phase I: Monitoring Perennial Pepperweed: Biology and Invasion

• Map occurrences of perennial pepperweed
• Investigate growth patterns and new patch establishment
• Gather background datasets (including LiDAR, and Virtual Herbarium data)

Phase II: Perennial Pepperweed Eradication and Ecological Monitoring

• Develop experimental eradication treatments
• Monitor the effects of proposed eradication treatments on non-target species
• Investigate soil chemical and physical properties in relation to perennial pepperweed
• Investigate the potential pre-emergent effects of the herbicide chlorsulfuron
• Monitor the species in the soil seed bank to establish herbicide effects and the active restoration potential for experimental eradication treatments

Phase III: Adaptive Management and Modeling

• Design an adaptive management framework that integrates the results of the study and can be used by Preserve managers and others interested in weed management.
• Model perennial pepperweed growth and new patch establishment

The below sections provide detailed recommendations and results for each of the Tasks and Subtasks completed between 2005 and 2011. Full reports are available online at http://baydelta.ucdavis.edu/pepperweed/.

1.1. Phase I: Monitoring Perennial Pepperweed: Biology and Invasion

Invasion resistance is linked to a multitude of biotic and abiotic factors including biodiversity and ecosystem function, but the lack of sufficient observational data has created uncertainties as to the role that many of these factors play. The debate centers on the question of whether natural areas with high native species richness and cover are immune to invasion by non-native species (Tilman 1999, Stohlgren et al. 2003, Fridley et al. 2007). The major barrier to resolving this issue is that different trends emerge when data is collected and analyzed at broad or fine spatial scales (Fridley et al. 2007). Data collected at broad spatial scales (>30m²) contends that these areas are highly invasible because they contain rare or endemic species which are extremely sensitive to disturbance (Stohlgren et al. 2003). Additionally, areas with high numbers of native species are commonly associated with high soil fertility, which can be easily colonized by an opportunistic invader. It is suggested that these sensitive sites may become prone to invasion if natural disturbances are altered or ecosystems processes are changed (Stohlgren et al. 1999, Davis et al. 2000). Alternatively, at a fine scale (<1m²), areas that maintain diverse functional groups, or have a heterogeneous mix of species that occupy the same habitat but use different resources, will be less invasible by occupying all available habitat niches. In the absence of anthropogenic disturbances, areas high in species richness and abundance will maintain functional group diversity, and by maximizing available resource use, will inhibit nonlocal invaders (Tilman 1997, Pokorny et al. 2005, Sheley et al. 2007). Perennial pepperweed is generally understood to multiply by its prolific underground root system, but Leninger & Foin (2009) provide evidence that the species exhibits high levels of seed production and viability at the Preserve (Leninger 2006, Renz and DiTomaso 2006). The rate at which seeds are produced at the Preserve provides strong evidence that the reported average of 3,231.5 seeds per inflorescence with a 96.4% germination rate will influence eradication success rates. In addition, Leninger & Foin (2009) found that seed viability only declines by 17% at the Preserve seven months after production and viable seed can travel up to five meters before there is a significant drop off in dispersed density. Their findings indicate that simply eradicating the plants will not wipe out a longstanding infestation and that closely monitoring treated sites will be instrumental in determining treatment success rates. As the main study sites tended to flood on a near-annual basis, it quickly became apparent that understanding not only the local hydrology but how perennial pepperweed and perennial pepperweed eradication methods might be affected by the hydrology would play a major role in the overall findings. Perennial pepperweed, like many other wetland or flood tolerant species, is capable of responding to flood conditions by altering its root to shoot ratio, producing adventitious roots, developing aerenchyma cells, decreasing photosynthesis, reallocating carbohydrate storage, and changing its nutrient uptake and allocation (Chen et al. 2002, Chen et al. 2005). By decreasing the root to shoot ratio, the species reduces its required oxygen, water, and nutrient uptake during stressful flood conditions (Naidoo and Naidoo 1992, Joly 1994), which are common in riparian and wetland areas at the Preserve. Adventitious roots, which grow beyond the flooded soil, can provide the species with oxygen necessary for survival in an anoxic environment (Jackson and Drew. 1984) and while adventitious root production is a common adaptation of flood tolerant species, it is not common in members of the Brassicaceae (Chen et al. 2002). The species has further adaptations which allow for the slowing of photosynthesis through a series of mechanisms such as stomatal closure and ethylene production resulting leaf senescence (Bradford 1983, Chen et al. 2002), this reduction is also linked to starch accumulation in leaf structures rather than root structures as inundated plants discontinue exchanging carbohydrates from shoot to root (Chen et al. 2005). Perennial pepperweed root structures increase soluble sugars when inundated which allows them to continue important processes like respiration without using their starch stores. The build-up of sugars during flood events may be linked to perennial pepperweed’s ability to recover after long periods of inundation (Chen et al. 2005). While inundated, perennial pepperweed’s nutrient concentrations of N, P, K and Zn decline while Fe and Mn increase (Chen et al. 2005). The adaptive ability of perennial pepperweed under short term flood conditions make the species an ideal invader on riparian floodplains and a threat to many ecologically unique habitats at the Preserve. To determine the scale and extent of the perennial pepperweed infestation at the Preserve, we surveyed over 1,371 acres over a period of four years using Trimble GPS units to define the total area infested. Our surveys identified over 1,100 individual patches (see Subtask 3.1). A subset of these populations, at what is known as the Experimental and Lower Floodplain, were revisited between 2005 and 2007, after an initial survey in 2004 (Cosumnes Research Group 2005). The populations that were revisited, between 312 and 456 patches a year, were used to determine annual patch growth rates and potential new patch establishment locations. We were able to determine the age of some populations using a root staining procedure, but concluded that this procedure was not useful at a large scale. New patch establishment and population growth was generally attributed to factors associated with the water year type, where wet years with long inundation periods produced fewer patches and dry years produced more patches. As the scale of the infestation was determined to be preserve-wide, we encourage eradication efforts to coincide with wet or dry growth type years, where treatment would be targeted in wet habitats during dry years and dry habitats during wet years. This treatment regime is further outlined in the adaptive management report (see Subtask 5.2).

1.2. Phase II: Perennial Pepperweed Eradication and Ecological Monitoring

Phase II was the most applied phase of the project, where patches were stratified by habitat type and randomly selected for an array of treatments and subsequent monitoring. The goal of this phase was to test a series of herbicide and non-chemical control methods, find a treatment that could be implemented with a high level of success, monitor each treatment’s effect on the overall ecosystem health, and ensure that a recommended treatment would not have a negative impact on existing Preserve resources.

1.2.1. Treatment Rationale

Our treatments consisted of using a combined method of mowing and applying glyphosate or chlorsulfuron to maximize the translocation of herbicide to the below ground root structures (Renz and DiTomaso 2004, 2006). We also included other experimental treatments for which little or no background information was available, including cut-stem herbicide treatments (Subtask 4.1), a tarping experiment (Subtask 4.2), and a weed pulling experiment. To examine how the impacts of all treatment combinations (Subtask 4.1; Subtask 4.2) would affect overall ecosystem health, we collected data on the pre-treatment and post-treatment non-target vegetation response (Subtask 4.3), soil chemical and physical properties (Subtask 4.4), soil residence of chlorsulfuron (Subtask 4.5), and the composition of the soil seed bank (Subtask 4.6) at multiple properties with different environmental conditions at the Cosumnes River Preserve. Treatment and subsequent monitoring took place over a period of five years (2005-2009) and included two consecutive years of treatment, one year of expanded treatment (herbicide only), and four years of monitoring to assess treatment efficacy, non-target vegetation response, soil impacts, and soil seed bank response. Mowing in combination with herbicide treatment increases success rates for some perennial species, including perennial pepperweed (Mislevy et al. 1999, Monteiro et al. 1999, Beck and Sebastian 2000, Renz and DiTomaso 2006). To better understand how phenological differences in mowed vs. un-mowed plots affected perennial pepperweed treatment success rates, Renz and DiTomaso (2004) tested the deposition, absorption, and translocation of applied glyphosate and analyzed root total nonstructural carbohydrates at various stages of perennial pepperweed phenology. These factors were tested by mowing experimental plots at the flower-bud stage or full flower stage followed by herbicide application at rosette, flower-bud and full flower stage. Experimental control plots were unmowed and treated with glyphosate at the flower-bud, full flowering and fruiting stage (Renz and DiTomaso 2004). As was hypothesized, Renz and DiTomaso (2004) found that glyphosate deposition was more abundant in plants with more above ground surface area. While mowing did not result in higher overall herbicide surface deposition compared to the control, a greater percentage of herbicide was deposited in the lower portion of the pepperweed canopy in mowed infestations. This study concluded that mowing decreased the re-sprouted height of a perennial pepperweed plant, and decreased the above ground sink for applied herbicide (Renz and DiTomaso 2004). To assess absorption and translocation of glyphosate in perennial pepperweed, Renz & DiTomaso (2004) selectively applied the herbicide to leaves in the flower-bud stage, the fruiting stage and a mowed flower-bud stage. Herbicide treated leaves were harvested after 48 hours and tested for glyphosate absorption. They found that mowed treatments absorbed more glyphosate, which was attributed to a less-developed cuticle in the leaves of the mowed plants. Translocation to root structures was assessed by harvesting roots of glyphosate-treated plants for analysis in mowed and un-mowed plants. As hypothesized, more glyphosate accumulated in the roots of mowed plants when compared to un-mowed plants. The roots of un-mowed plants contained very little herbicide translocated from the shoot to the roots. As observed in glyphosate deposition tests, perennial pepperweed generally re-sprouts to a shorter height after it is mowed allowing the herbicide to be applied in closer proximity to the roots. Root-total nonstructural carbohydrates were analyzed at different stages of perennial pepperweed phenology by harvesting roots in both mowed and un-mowed plants. Carbohydrate transport to the roots was greatest when plants were in the early stages of growth or after a mowing treatment. The authors suggest that increased carbohydrate transport to root structures in re-sprouting plants is evidence for the increased transport of glyphosate to roots in mowed plots (Renz and DiTomaso 2004). These findings support the claim that perennial pepperweed has a reduced ability to translocate herbicides to root structures from the upper canopy compared to the lower 3rd portion of its canopy. To test if mowing increases the efficacy of multiple herbicides Renz & DiTomaso (2006) tested chlorsulfuron, glyphosate, and 2, 4-D on pepperweed infestations. Chlorsulfuron was effective regardless of mow treatment, glyphosate increased in efficacy if plants were mowed first, but 2,4-D did not increase in efficacy after a mowing treatment. Both glyphosate and chlorsulfuron were identified as good choices for perennial pepperweed control. The authors note that glyphosate has low residence time in the soil, which is valuable when using plantings to revegetate a treated area (Renz and DiTomaso 2006). Because of the general importance placed on applying herbicide to the lower 3rd of the canopy to maximize the translocation of herbicide to the roots, we introduced an experimental cut stem, herbicide application method in an attempt to find a method with a lower herbicide application rate and one that could easily be conducted by volunteers. While this treatment style is generally applied to tree and shrub species of the woody variety, we hoped that the dense perennial root systems might react positively to this method. In addition, we implemented a previously understudied tarping method in an attempt to eradicate underground root structures without using an often objectionable herbicide method. Tarping reduces photosynthetic light and elevates soil temperatures. While annual species are commonly controlled using a tarp method (Horowitz et al. 1983), perennial species have larger root masses and are generally not controlled by tarping (Rubin and Benjamin 1984, Linke 1994). However, seeds generally do not survive when soil temperatures are elevated above 45° C (Horowitz et al. 1983, Rubin and Benjamin 1984, Peachey et al. 2001), which could reduce the potential for perennial pepperweed to reinvade after an eradication attempt.

1.2.2. Monitoring Rationale

Depending on the degree of infestation, the success rate of the removal, and the health of adjacent habitats, sites do not always respond favorably when relieved of a target species (Ogden and Rejmanek 2005 ). In areas where non-native species coexist within the native community, treatment-related disturbances may further degrade ecosystems by disrupting plant community dynamics, resulting in future invasions or other impacts (Rinella et al. 2009, DeMeester and Richter 2010). It is now widely accepted that declines in species diversity, functional group diversity, and vegetation density are factors that can contribute to a community’s invasibility (Ortega and Pearson 2005, Pokorny et al. 2005). These types of responses can also occur after invasive species removal and should be monitored to prevent future negative impacts. After an intensive herbicide treatment, the resultant plant community is more likely to be dominated by non-native annual grasses and other unaffected species thus reducing the functional diversity and increasing the invasibility of the area (Pokorny et al. 2005, Sheley and Denny 2006). Because of these factors, we monitored the non-target vegetation on an annual basis to look for potential negative and positive effects of different herbicide and non-chemical treatments. In general, some negative impacts were observed in both herbicide and tarp treatments. Herbicide use resulted in some conversion of native herbaceous understory to non-native annual grasses while tarp treatments resulted in large areas of bare soil, which are now open to future invasion (see Subtask 4.3). Re-vegetation may be necessary at the Preserve if negative results are observed in the response of the non-target vegetation or if the targeted species returns in years subsequent to removal. Properties at the Preserve with many dense patches of perennial pepperweed are likely to have large numbers of seeds stored in the soil seed bank. As it can be assumed that viable seeds will remain in the seed bank after most types of eradication, the outcome of some eradication methods, especially those in which bare soil is a common result, could result in new perennial pepperweed infestations. Perennial pepperweed seedlings are capable of outcompeting less aggressive seedlings (Spenst 2006) ultimately leading to ‘new’ infestations. In a system like the Preserve, passive restoration processes are commonly relied upon to establish more favorable vegetation following non-native plant eradication or a disturbance event. In areas with dense, pre-treatment perennial pepperweed cover, the composition of the seed bank will be partially responsible for ensuring successful passive restoration. In some cases, perennial pepperweed removal may be most successful when sites are actively restored and re-vegetated with competitive native plant species (Eiswerth et al. 2005). To that end, the diversity and abundance of species in the seed bank, as well as any effect the perennial pepperweed removal processes had on that seed bank and resulting non-target vegetation (Subtask 4.6) was evaluated. Soil underlying sites densely populated by perennial pepperweed was shown by Renz & DiTomaso (2004) to have higher levels of N, Ca, Mg and lower levels of acetate-extractible Na. Higher nitrogen levels in locations of perennial pepperweed growth were related to perennial pepperweed’s ability to reduce N-cleaving enzymes in the soil (Blank and Young 2002). The species has a greater concentration of calcium in its tissues than the average plant and may actually input Ca to the soil, increasing soil friability. This process facilitates amelioration of sodic soils as calcium and magnesium levels increase and replace sodium on clay exchange sites. During flood events, sodium leaches from the upper soil profile, as it was previously replaced by calcium and magnesium on clay exchange sites, (Renz and Blank 2004). Even though perennial pepperweed improves sodic soils by relieving salt from the clay, it creates saline soils via the decomposition of thatch and leaf litter. (Blank and Young 2002). We investigated the overall impact of perennial pepperweed on a number of soil physical and chemical properties and found some impact, which is detailed below and in Subtask 4.4. Both glyphosate and chlorsulfuron are effective at eradicating perennial pepperweed (Renz and DiTomaso 2006) but there is some evidence that chlorsulfuron and herbicide use in general, has a negative impact on non-target vegetation and can have a long residence time in the soil (Guo and Sun 2002, Young et al. 2002). Monitoring the effects of herbicide use on non-target vegetation is essential to an ecologically based weed management regime in order to determine an appropriate eradication method (Maxwell and Luschei 2005). Chlorsulfuron has a wide range of soil and pre-emergent effects, but is more of a broadleaf specific herbicide and does not affect grasses as strongly as it affects broadleaf herbaceous plants (DuPont 2003b). The variable residence time of chlorsulfuron depends on environmental factors such as soil temperature, moisture and pH (DuPont 2003b). Glyphosate, on the other hand, breaks down quickly in the soil and is not known to negatively affect soil physical and chemical properties (Dupont 2003a). Areas treated with glyphosate can be replanted or seeded soon after control, thus decreasing the likelihood of reinvasion (Renz and DiTomaso 2006). We tested the pre-emergent impact of chlorsulfuron on root growth and found that its long term effects (greater than 120 days) were minimal (Subtask 4.5). While this finding is interesting and could be used as an argument for continuing to use chlorsulfuron at the Preserve, its use is now restricted to properties that are not adjacent to waterways, significantly reducing its application potential. By identifying barriers to eradication success, such as residual herbicide effects, seed bank depletion, bare soil exposure, low adjacent species and functional group diversity, infested areas can be treated with more sensitivity and a whole ecosystem approach can be developed (Davis et al. 2000, Guo and Sun 2002, Ogden and Rejmanek 2005).

1.3. Phase III: Adaptive Management and Modeling

Phase III, as implemented by Subtask 5.1 and Subtask 5.2, was the culmination of this project wherein the results from all of the other tasks were used to develop the adaptive management framework and the perennial pepperweed growth models. While these two tasks are distinct and very dissimilar, their completion required integration of the results from all the other tasks. We chose to use multiple modeling approaches in order to capture perennial pepperweed’s growth and establishment at the Preserve. Once all relevant data were collected, including five years of population data, USGS river gage data at Michigan Bar, elevation, canopy cover, canopy height, inundation, experimental control efficacy, among others, we ran three complementary models: dispersion, physical drivers, and rapid detection. The models each predicted growth a little differently, but all predicted that new patches could establish at many of the Preserve’s properties in a number of sensitive habitats. The prediction surfaces create population growth probabilities based on the correlation of environmental variables. These models provide detailed growth predictions and are explained in the sections below and in the Subtask 5.2 report. The Preserve can utilize these prediction surfaces to assist them in making restoration decisions and in allocating weed control resources. To address population growth we modified an existing Leslie Matrix and Lefkovich Matrix Population Model for teasel (Dipsacus sp.) to address how experimental treatments would impact different rates of growth. This population model relied on life history traits, such as seed production and germination rates, to model individual patch growth through time. We adjusted growth rates by calculating patch growth for different years (data collected in Subtask 3.1) and establishing those as dry year, wet year, and moderate year or stable growth. The results from this population model informed our adaptive management framework buy providing us with a treatment repeat rate that would appropriately manage the species.