When implemented correctly, monitoring plays a crucial role in ecological restoration. A well-designed monitoring program can determine whether a project was implemented as planned and whether objectives are being met, as well as build public support and improve project efficiency through adaptive management. Despite these and other valuable outcomes, monitoring is rarely implemented to the necessary extent, especially with respect to changes in vegetation. One reason for this is that conventional approaches to vegetation monitoring are time-consuming and achieving necessary levels of precision can be expensive. Qualitative and remote sensing methods have been proposed to be more cost-effective. For example, managers are using a rapid monitoring method, the Qualitative Rapid Assessment (QRA), to assess river and floodplain responses to restoration treatments on the Clark Fork River in Montana, to reduce costs of data collection. Remote sensing using UAV (unmanned aerial vehicle) technology is also becoming a common method to monitor vegetation due to the ability to provide high resolution imagery for large areas quickly with minimal labor costs. However, how well qualitative and remote sensing data compare to on-the-ground quantitative data needs to be further investigated. Our proposed project aims to compare on-the-ground quantitative data to data collected using the QRA and interpreted from high resolution imagery for monitoring vegetation cover. Specifically, we will compare mean percent cover and calculate and compare measurements of observer error, margin of error achieved, and required sample sizes among the three types of monitoring for three metrics: woody vegetation cover on streambanks, woody vegetation cover in the floodplain, and herbaceous vegetation cover. Based on our findings, we will recommend modifications to QRA and remote sensing protocols and sampling design to improve both precision and efficiency. Our findings will contribute to the effective implementation of monitoring at the largest superfund complex in the United States – the Clark Fork River Superfund Complex – and to improving monitoring methods for use in other restoration projects.

Radon 222 (222 Rn) was measured along 8.7 kilometers of the Clark Fork River, between Harper’s Bridge and Frenchtown, MT. 12 water samples were taken along the stretch. Samples 1 through 4 and 10 through 12 were collected on a 1 km interval, samples 5 through 9 were taken on a 500 meter interval. Samples were analysed for dissolved 222 Rn using a RAD7 spectral alpha decay detector. Instream 222Rn was modeled to quantify groundwater discharge to the river. Well logs from the Montana Groundwater Information Center and literature on the Missoula Valley aquifer were analyzed. Analysis of well logs adjacent to the river reveals an alluvial aquifer system to the east consisting of interbedded gravel, sand, silt, and clay. To the west, bedrock rises steeply from underneath the river to crop out at the surface. Analysis of the samples reveals that there are measurable quantities of 222 Rn through the entire stretch sampled, starting at 395 mBq/L near Harper’s bridge, with peaks of 950 mBq/L at 2 km and 632 mBq/L at 6.5 km. Lowest concentrations were 395 mBq/L at the start of sampling, 355 mBq/L at 4.3 km, and 336 mBq/L at 8.7 km. Modeling results averaged to 5.5*105 m3/day of groundwater entering the river, with a standard deviation of 1.2*105 m3/day, occuring in areas of high 222Rn activity. This work identifies and quantifies the spatial distribution of groundwater discharge at the west end of the Missoula Valley postulated by previous works.

Reference models are critical to ecological restoration because they allow managers to understand the condition the project site would have been in if degradation had not occurred, to develop and communicate a shared vision of project targets, and serve as the building blocks for developing treatment plans. However, despite their importance, many restoration projects do not include information on reference conditions or include data from only a single or limited number of reference sites. We are proposing to create a reference model that can be used to evaluate riparian vegetation restoration after the removal of a dam on Rattlesnake Creek (Missoula, Montana). Specifically, we will identify streams that are environmentally similar to Rattlesnake Creek but have no to minimal degradation and look at the naturally occurring vegetation on Rattlesnake Creek at several undisturbed locations. At each of these “reference sites”, we will collect data on percent cover of vegetation adjacent to the stream. We will use the vegetation data to calculate mean reference conditions for common species composition and structure. In addition, we will conduct power analyses to determine the precision of our estimates, as well the number of sites that would need to be included in the model to achieve the desired level of precision. After data analysis, we will present our findings to the community for consideration for guiding revegetation in the area in which the dam is removed. Our proposed project will contribute to restoration success on Rattlesnake Creek, as well as increase Missoula residents’ excitement and knowledge about restoration. In addition, our results can be used as a guide for developing reference models for future dam removals or other restoration projects throughout the western US.

Like many streams in the northern Rocky Mountains, Ninemile Creek in Western Montana was degraded by placer mining. This type of gold mining leaves a legacy of physical transformation to the stream, characterized by a highly incised and straightened river channel disconnected from the floodplain. These changes to the physical structure cause low streamflow during much of the year which is then punctuated by increased pulses of water during spring snowmelt and even rainfall events. This combination of physical and hydrologic changes render the creek an unsuitable habitat for many invertebrates, fish species, and mammals, including beavers. Recently, restoration in the Ninemile Creek watershed led by Trout Unlimited (TU) has reintroduced sinuosity and increased floodplain interaction with the river channel in three reaches of the stream which were restored in 2014, 2016, and 2018. To expedite recovery of the biotic component of this newly constructed floodplain ecosystem, TU planted willows and used a native seed mix representative of riparian plant assemblages in the region. To evaluate the riparian vegetation recovery, soil characteristics, and beaver presence, a post-restoration monitoring effort was implemented in 2019 across all three restored reaches. Line point intercept (LPI) method was used to assess vegetation growth forms and surface types along transects perpendicular to the stream channel, beginning at the bank. Signs of moose and deer browse were also documented when observed within the transect. Soil samples were collected at a single randomly selected point on each transect and beaver structures were documented and mapped along the restored reaches. These measurements were taken at each restoration site, and at a reference reach which never experienced placer mining. The results show measurable differences in percent cover and distribution of vegetation types, predominant soil characteristics and beaver presence among the reaches. The data gathered will inform TU on how to more effectively restore and adaptively manage their current and future restoration activities on the Ninemile Creek and other similar restoration projects.

In graph theory a Hamilton cycle is a walk around the vertices of a graph in which each vertex is visited exactly once, and then it returns to the starting vertex. The problem of determining whether a graph contains a Hamilton cycle has been studied extensively and is determined to belong to the so-called NP-complete family of problems for arbitrary graphs. Due to the difficulty in solving such a problem for an arbitrary graph, we set our sights on a family of graphs described by graph theorist John Sheehan. A maximum uniquely Hamiltonian graph contains the greatest number of edges possible while maintaining a single Hamilton cycle. Sheehan shows that for a graph with n nodes (for n >= 4), the maximum number of edges it can contain is equal to (n^2/4) + 1. We will describe an algorithm that finds the Hamilton cycle for any such graph or any of its subgraphs in polynomial time. This algorithm shows that these graphs do not suffer the same complexity issues as do arbitrary graphs for a Hamilton cycle problem. For any graph containing a single Hamilton cycle, that cycle can be revealed in polynomial time.