FACTORS AFFECTING ECOSYSTEM RECOVERY AFTER PLACER MINING IN NORTHWESTERN BRITISH COLUMBIA by Jose Haig BES, York University, 2014 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIRMENTS FOR THE DEGREE OF MASTER OF NATURAL RESOURCES AND ENVIRONMENTAL STUDIES UNIVERSITY OF NORTHERN BRITISH COLUMBIA January 2017 © Jose Haig, 2017 ii Abstract This research sought to determine the attributes affecting unassisted ecosystem recovery after placer mining. Ninety post-mining sites in 14 creek drainages east of Atlin Lake, British Columbia, were sampled to represent a range of times since disturbance (9 to 76 years). The six indicators utilized (vascular species richness, plant community similarity to undisturbed reference sites, summed plant cover, structural diversity, A-horizon depth, wildlife activity) exhibited different recovery trajectories and dependencies. Across all six indicators, the factors most important to ecosystem recovery were, in order of importance: time since disturbance, microsite relief, elevation, slope position, and soil texture. Without any reclamation, linear extrapolation indicates that a mean of 101 years would be needed for disturbed sites to return to mean undisturbed conditions. Classification and regression tree analysis identified thresholds of these factors that may promote or hinder recovery. These thresholds were used to refine recommendations for promoting ecosystem recovery after mining. iii Acknowledgements This project was funded by the British Columbia Ministry of Forests and Natural Resource Operations (FLNRO) Research Program, with preliminary site inspections and follow-up soil analysis funded by the Taku River Tlingit First Nation. Field assistance was provided by Rick Trowbridge, Carla Burton, and Clive Aspinall. Considerable local and disciplinary insights were provided by Mark Connor, Chris Apps, Eric Telford, Nicole Gordon and Anna Schmidt (of the Taku River Tlingit Land and Resources Office), Eric Smith and Karen Diemert (FLNRO), Anne Moody, Kathie Wagar, and Doug Flynn (B.C. Ministry of Energy and Mines), Doug Hall and several individual working miners (Atlin Placer Miners Association), and Clive Aspinall (geologist and long-time community member). Laboratory facilities were provided by the Enhanced Forestry Lab at the University of Northern British Columbia, with additional assistance and advice provided by Doug Thompson and Benita Kaytor. Thanks to Kishari Arachchilage of Western Ag Innovations Inc. for assistance in interpreting soil nutrient data. Thanks to Darin Brooks (College of the North Atlantic) for his assistance in getting started with the CART analysis in R. Special thanks to members of my supervisory committee -- Phil Burton, Roy Rea and Dave Wilford -- for their advice and ongoing support. iv Table of Contents Abstract .............................................................................................................................................................. ii Acknowledgments........................................................................................................................................ iii Table of contents .......................................................................................................................................... iv List of tables .................................................................................................................................................... vi List of figures ................................................................................................................................................. vii Chapter 1. Introduction .............................................................................................................................. 1 Relevance of research .................................................................................................................. 3 Chapter 2. Literature review.................................................................................................................... 6 Factors affecting ecosystem recovery .................................................................................. 6 Current best management practices ...................................................................................12 Chapter 3. Methodology ...........................................................................................................................15 Laboratory methods ...................................................................................................................25 Statistical methods ..................................................................................................................... 28 Chapter 4. Results .......................................................................................................................................31 Disturbed conditions ..................................................................................................................32 Regression analysis .....................................................................................................................34 Classification and regression tree (CART) analysis .....................................................38 Chapter 5. Discussion ................................................................................................................................43 Chapter 6. Conclusions and recommendations ............................................................................53 Suggestions for future research ............................................................................................55 References cited ...........................................................................................................................................58 v Appendices Appendix I. Field data collection form. .............................................................................................65 Appendix II. Site and vegetation data used in statistical analysis. ......................................67 Appendix III. Plant species frequency (%) on areas disturbed by placer mining (n=90) and nearby undisturbed (n=10) “control” sites east of Atlin Lake, B.C. .......................... 71 Appendix IV Additional classification and regression trees. ..................................................74 Appendix V. Significant (p<0.05) differences in ecosystem recovery indicators among categorical subsets of the data, based on ANOVA results. .......................................................78 Appendix VI. Significant (p<0.05) correlations of site, soil, and vegetation factors with indicators of ecosystem recovery. .......................................................................................................79 vi List of Tables Table 1. Comparison of best management practices by jurisdiction .................................14 Table 2. Estimation of average offset between tree or shrub ages and evidence of the last mining or reclamation activity .....................................................................................................26 Table 3. Representative condition on undisturbed sites in Atlin area (n=10) ..............31 Table 4. Compilation of wildlife activity evidence presence found on sites, for both disturbed and undisturbed sites. .........................................................................................................34 Table 5. Significant (p<0.05) regressions of indicators on estimated time since disturbance, and projected recovery times to alternative thresholds ..............................37 Table 6. Compilation of ranked importance of attributes on six indicators of post-mining ecosystem recovery ....................................................................................................................................42 vii List of Figures Figure 1. Former mining sites sampled east of Atlin Lake in northwestern British Columbia. All sites are associated with stream systems and have road access. ...........18 Figure 2. Depiction of 30m transect laid over a sample plot (11.28m diameter) .......20 Figure 3. Depiction of subplots within main plot. Four subplots were utilized one at the center of the plot, one located at the northern-most perimeter of the main plot, one southeast at 120 degrees’ azimuth from north, and one located southwest at 240 degrees’ azimuth to north .......................................................................................................................24 Figure 4. Plot 15-01 demonstrates a site with sparse vegetation and exposed bare mineral soil in many area, the estimated time since disturbance is ten years. .............32 Figure 5. Plot 15-10 demonstrates a plot with an established tree and shrub layer and minimal bare mineral soil exposure. The estimated time since disturbance is 45 years. ..............................................................................................................................................................................33 Figure 6. Summed plant cover (sum of tree, shrub, herb and moss/lichen layers) as related to time since disturbance estimated from tree or shrub annual ring counts at placer mined sites east of Atlin Lake, BC. ........................................................................................35 Figure 7. A horizon depth as related to estimated time since disturbance after placer mining ...............................................................................................................................................................36 Figure 8. Classification and regression tree for factors influencing vascular plant richness (species per 100 m2) on post-mining sites east of Atlin Lake, B.C. R 2 = 0.39 ..............................................................................................................................................................................40 Figure 9. Classification and regression tree for factors influencing summed vegetation cover (%) on post-mining sites east of Atlin Lake, B.C. R2 = 0.41. .......................................41 viii Figure 10. Soil profile in Plot 15-05, showing negligible soil development; estimated time since disturbance is 20 years ...............................................................................49 Figure 11. Soil profile in Plot 20-08, showing an apparently well distinguished A horizon (to 9 cm), and B horizon (to 20 cm); estimated time since disturbance is 33 years. .................................................................................................................................................................49 Figure IV-1. Classification and regression tree for factors influencing compositional similarity to control sites on post-mining sites east of Atlin Lake, B.C. R2 = 0.46. .......74 Figure IV-2. Classification and regression tree for factors influencing structural diversity (H’) on post-mining sites east of Atlin Lake, B.C. R 2 = 0.51. ................................75 Figure IV-3. Classification and regression tree for factors influencing A horizon depth (cm) on post-mining sites east of Atlin Lake, B.C. R2 = 0.30. ..................................................76 Figure IV-4. Classification and regression tree for factors influencing wildlife species on post-mining sites east of Atlin Lake, B.C. R2 = 0.63. ....................................................................77 1 Chapter 1. Introduction Placer gold mining has been ongoing in northwestern British Columbia for over a century, particularly in the area immediately east of Atlin Lake (Dickinson & Smith, 1995). Sparked as an offshoot of the Klondike Gold Rush (1896–1899), placer gold mining began in the Atlin region in 1898 (Dickinson & Smith, 1995) and by the end of the year over 3000 gold miners had flocked to streams in the Atlin area. In that same year, the production of gold was estimated to be more than 3,750 ounces, and by 1899 more than 400,000 ounces of gold had been recovered from the area (Griffin, 1992). Today, Atlin is a small community nestled in the northwestern corner of British Columbia, resting on the traditional lands of the Taku River Tlingit First Nation. Atlin currently has a population of approximately 400 full-time residents (Statistics Canada, 2011). However, during the warmer months, the population balloons to approximately 1,000 to 1,500 residents; the many summer residents include artists, retirees, tourism workers, and part-time placer miners. Placer mining is currently ongoing in the Atlin area, under modern environmental standards, regulations, and land use agreements (BCTRTFN, 2011). All active mines also conform to practices required in special management areas, and protocols to ensure shared decision making with First Nations (TRTFNBC, 2011). In particular, the Blue Canyon / At Xá Koogu area as outlined in the Wóoshtin wudidaa: Atlin Taku Land Use Plan is classified as an area-specific management zone designated with the aim of reducing the levels of net land disturbance (BCTRTFN, 2011). As interpreted by the Blue Canyon Technical Subgroup, this means that some disturbed locations must transition to a ‘recovered’ or ‘target’ state before any new mining 2 operations may proceed in the area (BCTS, n.d.). Criteria for when a location has reached a ‘recovered’ state remains to be defined. Placer gold mining typically disturbs the configuration of a landscape and thus the ecological processes at work there through its very nature. Placer mining is a resource extraction method that targets alluvial deposits, also known as placers. Originally worn from mineral veins in the bedrock, gold dust, flakes, and nuggets have settled into alluvial sands and gravels, due to the movement of water. These alluvial or placer deposits are found buried underground (but above bedrock) and also in streambeds. Placer mining requires removal of soils and gravel (overburden) to gain access to the valuable minerals found in the placer deposits (LaPerriere & Reynolds, 1997). As in the Atlin area, placer gold mining also has a long history in the neighboring Yukon territory, as well as the interior of Alaska, USA, and elsewhere around the world. Jurisdictions around the world vary with regard to the degree of mining regulation and the amount of active reclamation work required after a placer mining operation has concluded. As placer mining operations are generally concentrated around stream channels, the distance to an undisturbed ecosystem is often short, with natural processes often rapidly revegetating the location. In northern British Columbia, as well as elsewhere, natural plant succession has commonly been entrusted to revegetate formerly mined sites, with little to no investment in active revegetation or reclamation required (TRTFN, 2013). The goal of the research described in this thesis is to determine the attributes affecting ecosystem recovery after placer mining. The near-continuous nature of 3 mining, accompanied with relatively unmanaged revegetation in the area, has created opportunities in the Atlin area for estimating the time needed for ecosystem recovery (BCTRTFN, 2011). Given the management intent of decreasing the net area in a disturbed state (BCTRTFN, 2011), it is also the goal of this research to identify a default time to recovery for abandoned mine spoils, and to identify the factors or practices that may reduce or prolong the time needed to meet certain thresholds. Relevance of research Governments and interested parties associated with the greater Atlin community have many varying interests and values, including those of the Taku River Tlingit First Nation, the Atlin Placer Miners Association, and the government of British Columbia. Recognizing the different values held, the Wóoshtin wudidaa: Atlin Taku Land Use Plan (BCTRTFN, 2011) allows for the interests of multiple parties to be considered with some compromise in the Blue Canyon / At Xá Koogu area. Given the extensive history of placer mining in the Atlin area, it is recognized that placer mining holds a key role in the local economy and in defining its cultural identity. Continuous access to mining, prospecting, and tourism opportunities are essential to ensuring the economic sustainability of the Atlin area. In order to halt further degradation of natural ecosystems, the Wóoshtin wudidaa: Atlin Taku Land Use Plan requires disturbed sites to return to a ‘recovered’ state before an expansion of placer mining operations can proceed (BCTS, n.d.). The requirement also allows placer miners to continue operations at current levels, supporting economic viability within the area while simultaneously preventing an 4 expansion in the net area of disturbed land. Reducing the time needed to return disturbed land to its target state carries multiple benefits: it not only reduces the time mining operations must wait before beginning a new operation, but it also benefits the broader ecological landscape by increasing the abundance of ecosystems dominated by natural species compositions and ecological processes, minimizing the overall extent of disturbed and degraded land on the landscape. The Wóoshtin wudidaa: Atlin Taku Land Use Plan offers a unique view into the sustainability of the Atlin area, both economically and environmentally. With the historical reliance on placer mining, the sustainability and economic viability of the Atlin community will eventually hinge on balancing environmental issues with economics (Alavalapati & Adamowicz, 2000; Hayter, 2000). The introduction of the Wóoshtin wudidaa: Atlin Taku Land Use Plan is able to create a paradigm shift within the Atlin area. Considering that ecosystem recovery rates are affecting placer mining operations, having “best management practices” (BMPs) in place for mine closure plays a vital role in ensuring economic sustainability. With the inability to raise the net disturbance of an area, it becomes beneficial for mine operators to pursue accelerated ecosystem recovery rates on sites previously disturbed or currently being mined. In order to sustain operations in the Atlin area, given the finite level of active mining permitted, mine operators must ensure that ecosystem recovery is occurring and it is to their benefit to lower the amount of net disturbance in order to continue or expand operations. As such, maximizing ecosystem recovery rates becomes of interest to those who otherwise may have had no interest in ecosystem 5 recovery or ecological values prior to the introduction of the Wóoshtin wudidaa: Atlin Taku Land Use Plan. This mix of changing views and the use of area-specific policy allows for a unique view on the Atlin area and accelerated ecosystem recovery within a northern Canadian context. With ecosystem recovery rates being of interest to multiple parties in the Atlin region, ecological factors affecting ecosystem recovery should be examined and well represented in BMPs to be recommended for implementation over the course of mine operations and closure. Given the variability in site conditions and the attributes of flora and fauna in the region, it is clear that many ecological and topographical factors can have an immense effect on the rate of ecosystem recovery (Entrix, 1986; Polster, 2011). As such, the literature review and description of original research that follows examines multiple ecological, topographical, and management considerations that can affect ecosystem recovery and could be incorporated in future BMPs. 6 Chapter 2. Literature review There have been various studies conducted after placer mining operations have halted, ranging geographically and encompassing various objectives. Many studies focused on the environmental impacts placer mining may have on surrounding ecosystems, and possible management techniques to better promote natural succession and ecosystem recovery (Farrington, 2000). Numerous studies focused on the early stages of ecological development following mining (Densmore, 1994; Prach & Hobbs, 2008). Many of the studies reviewed utilize a ‘chronosequence’ or space-for-time sampling technique as an alternative to long-term study; this method allows for sites of different ages but similar attributes to be sampled (Pickett, 1989). Similar to long-term studies, chronoseqeunce-based studies are utilized to detect multi-decadal trends. Factors affecting ecosystem recovery Several authors point out the importance of ameliorating harsh abiotic conditions (e.g., severe temperatures, moisture and nutrient limitations) in the early stages of primary succession and mine reclamation (Bradshaw & Chadwick, 1980; Johnson, 1987; Gretarsdottir et al., 2004). Over longer time periods, biotic interactions must be considered and may need to be manipulated (e.g., through the removal of exotic species or brush control) in order to direct post-disturbance vegetation to its desired target state (Walker et al., 2007; Polster, 2011). Substrates consisting primarily of high amounts of gravel and cobbles – the predominant residue from the processing of placer deposits – have a limited ability to retain water and nutrients, supporting large and numerous interstitial spaces with little cation exchange capacity. Consequently, plant growth, nutrient cycling and other 7 ecosystem processes are constrained on such coarse and barren soils (Russell, 1973). Processed mine tailings from the placer gold mining industry are particularly devoid of soil material (with coarse fragment content often >70%), because the clays, silts, and even sands are washed out of the matrix in efforts to extract the gold. Consequently, mine reclamation and the promotion of ecosystem recovery through natural succession or artificial revegetation first depends on conditions or techniques that encourage the retention of fine soil materials, water and nutrients (Bradshaw & Chadwick, 1980). Studies focused on reclaiming gravel sites in a northern context have highlighted the importance of soil reconstruction (Entrix, 1986; Johnson, 1987; Farrington, 2000). Adding a topsoil cap can mitigate many of issues that barren mine sites pose to ecosystem recovery. In addition to directly supporting plant growth, placing a cap of topsoil on abandoned placer mine sites offers multiple benefits, including the improvement of water retention, organic matter content, and decrease the rate of erosion (Johnson, 1987; Christensen et al., 2013). For example, a 5-cm topsoil cap was found to promote significantly accelerated plant establishment and recruitment on abandoned placer mine tailings in interior Alaska compared to sites where no topsoil cap was placed (Cooper & Van Haveren, 1994). Plant establishment success and plant growth rates were seen to increase as the amount of topsoil increased from 0 cm to 5 cm in depth; similarly, water retention also increased with more topsoil (Cooper & Van Haveren, 1994; Christensen et al., 2013). Furthermore, it was found that vegetation cover increased faster from a barren state when topsoil had been mixed with the barren parent material rather than utilizing a topsoil cap, and vegetation cover returned faster when lower soil horizons were also reconstructed (Halvorson et al., 1986). Ideally, 8 post-mining sites that are dominated by gravel and cobbles should be reconstructed by utilizing stockpiled topsoil to replace material previously removed. However, if topsoil is unavailable (which is often the case when reworking former tailings), the introduction of other fine materials can be used to accelerate ecosystem recovery (Entrix, 1986; Johnson, 1987). Re-contouring of placer mining sites after operations have ceased can have multiple benefits to future ecosystem recovery. Re-contouring also allows for accelerated ecological succession as it removes extreme slopes and limits erosion of fine soil materials (Entrix, 1986; Densmore, 1994). Extreme slopes are not only likely to exacerbate the rapid loss of water, but that water will also remove valuable fine material, organic matter, nutrients, and plant seeds. But if improperly done, recontouring operations can result in accelerated erosion and/or compaction of the substrate. As such, any machine traffic should be minimal and tracks and furrows should follow the elevation contours, and not track up and down slopes (Polster, 2011). The roughening of surface topography during the process of site re-contouring can also accelerate ecosystem recovery in terms of initial plant establishment. Smooth surfaces are unable to capture seeds and support seedling establishment to the same degree as rough surfaces with varying evenness (Evans & Young, 1987). A roughened surface compared to a smooth surface was found to be able to better capture and retain organic matter, improve water retention and better capture seeds (Evans & Young, 1987). A rough surface also aids in preventing soil erosion via rainfall and overland flow: a rough surface in comparison to a smooth surface was found to slow water 9 movement across the surface and as a result was found to have drastically lower rates of soil erosion (Römkens et al. 2002). The replacement of overburden onto abandoned sites significantly accelerates the rate of ecosystem recovery by allowing a greater array of plant species to establish. Where no soil material is present, the substrate is often coarse or compacted, and accompanied with a harsh climate; in these conditions, it may take centuries for soil to develop and for vegetation to establish (Polster, 2011). Recovery times to a target ecosystem vary depending on geographic location, site location on the floodplain, and the target ecosystem selected. Without active ecosystem enhancement, former placer mining sites located in the Klondike were found to take approximately 20 years to establish shrub thickets/communities, and full recovery to a coniferous forest required over 80 to 100 years (Entrix, 1986). The use of reclamation practices aimed at accelerating ecosystem recovery can reduce the amount of time needed for a shrub thicket/community to establish to less than ten years from initiation of the reclamation process (Entrix, 1986; Densmore, 2005). Similarly, topsoil replacement and tree planting has resulted in successful plant community reconstruction within 12 to 15 years of oil shale mining in Estonia, although floristic richness, soil nutrients and organic matter accumulation continue to increase over time and differ between sites planted to birch and sites planted to pine (Laarmann, 2015). The removal of overburden to conduct placer mining removes fertile soil and organic matter present prior to disturbance. The stockpiling, protection, and replacement of overburden onto placer mining sites is outlined in the recommended management practices of many jurisdictions, and has been observed to accelerate 10 ecosystem recovery after placer mining (Entrix, 1986; Densmore, 1994; Farrington, 2000). Viable propagules (seeds, spores, rhizomes) of plants removed as part of premine site clearing are often found in the topsoil and can remain viable for some time. As found by Miller et al. (1985) there is a significant reduction of viable plant propagules after two years of topsoil storage time. If possible, such material should be stored with overburden in a manner that protects the plant materials and soil from further degradation (e.g., through exposure to anaerobic conditions) or erosion, thereby enabling a more rapid recovery of vegetation once overburden is re-spread on the placer mining sites (Entrix, 1986; Densmore, 1994; Farrington, 2000). There is concern that long-term storage of topsoil may result in the deterioration of seeds, rhizomes, and important rhizosphere bacteria and fungi (Thurber Consultants et al., 1990; Miller & Jastrow, 1992). Consequently, soil storage times should be kept to a minimum (less than two years) in order to protect the viability of these and other important ecosystem components (Thurber Consultants et al. 1990; Sheoran et al. 2010). Limiting soil erosion assists in accelerating ecosystem recovery by preserving fine materials and organic matter found in the growing medium and helps prevent further downstream disruption caused by the increased amount of sediment within stream channels (Entrix, 1986; Densmore, 1994). Preventing soil erosion is essential to accelerating ecosystem recovery rates, especially where substrates (such as placer tailings) are coarse (Johnson, 1987; Farrington, 2000). Soil erosion detracts from a site’s ability to retain water, retain nutrients, and support plant life. Multiple methods have been mentioned to protect reclaimed sites from soil erosion, including light soil 11 compaction, introduction of plant life to bind the soil, the use of hay or straw as mulch, erosion blankets, and fabric erosion barriers (Entrix, 1986; Farrington, 2000; SMEGAC, 2012;(BCMFLNOEM, 2014). Efforts towards the alteration of barren gravel sites to include topsoil would be futile without measures in place to prevent future soil erosion from occurring. Soil fertility and the presence of organic matter affects plant growth and consequently ecosystem recovery rates. Encouraging vegetation and plant litter accumulation not only helps prevent soil erosion (Pimentel & Kounang, 1998), but over time and through the process of decomposition adds additional organic matter, improved moisture retention and nutrient retention capacity to the soil. Even with no topsoil present and limited organic matter content, soil fertility can be increased by certain plant species, particularly nitrogen-fixing plants that are symbiotic with Rhizobium (legumes, family Fabaceae) or Frankia (such as Alnus, Dryas, and Shepherdia species) microorganisms. Dryas and Alnus species are known to accelerate forest succession on gravel bars in the Tanana River of central Alaska (Walker et al., 1986). Similarly, Shepherdia canadensis (L.) Nutt. has been found to boost soil nitrogen content in floodplain substrates (Rhoades et al., 2008). In a ten-year study in Alaska, Alnus viridis (Chasix) DC was planted on abandoned placer mining sites, where it facilitated improved growth and vigour of associated Salix alaxensis (Andersson) Coville and Populus balsamifera L. plants (Densmore, 2005). Rather than beginning the reclamation process after the closure of a placer mining operation has occurred, the option of progressive reclamation offers some benefits to ecosystem restoration. Progressive reclamation means that reclamation 12 efforts (i.e., clean up, slope stabilization, topsoil spreading, etc.) would occur concurrently with mine operations, throughout and after the lifespan of mining operations. Reclamation efforts occurring concurrently with mine operations allow for further acceleration of the reclamation process. Progressive reclamation can benefit parties interested in ecosystem recovery and mine operators simultaneously, as ongoing reclamation during the lifespan of a mine site could “fast-track” ecosystem recovery and optimally shave years from recovery times compared to sites that are not utilizing progressive reclamation. Secondly, progressive reclamation spreads the cost and workload required for reclamation throughout the lifespan of the mine, rather than just at mine closure, resulting in a reduced net cost and workload at the time of mine closure compared to sites not utilizing progressing reclamation (Green et al., 1992; NAIT, 2015). The conservation of viable plant propagules, rhizobacteria, and fungal symbionts through reduced topsoil storage times is another reason why progressive reclamation and development is encouraged (Green et al., 1992). Current best management practices Many legal reclamation requirements and recommendations from different political jurisdictions suggest similar practices in order to promote accelerated ecosystem recovery after placer mining (Table 1). Given the fact that placer mine tailings and excavation areas share many characteristics with gravel pits and quarries, guidelines for their reclamation (e.g., Green et al. 1992) provide useful advice as well. Common recommendations include: a pre-disturbance survey and site assessment to characterize baseline and target ecological conditions prior to any mining operation; the storage and protection of plant materials and overburden removed during the 13 mining process; re-contouring of sites after operations have ceased; and the replacement of overburden and original species removed from the site (Entrix, 1986; Densmore, 1994; Farrington, 2000; Rieger et al. 2014). 14 Table 1. Comparison of best management practices by jurisdiction. Atlin Area, BC, Canada Alaska, USA Saskatchewan, Canada Yukon, Canada Idaho, USA Mongolia United States Agency for (BCMFLNOEM, 2014) (Entrix, 1986) (SMEGAC, 2012) (Yukon Tourism and (Idaho Department of (Farrington, 2000) International Development Culture, 2010; Yukon Lands, n.d.) Practice (Hagler Bailly Inc., 1998) Government, 2013) Re-contouring of mine -Slope and tailing -Recommended as -Not Required, tailings gradation required, soon as possible Topsoil replaced - Not Required - Sites should be re- -Not Required - Recommended to be recommended when contoured to - Erosion prevention based on surrounding soil along with surface high rates of erosion are approximate previous methods and site characteristics roughening foreseeable contour recommended -Replacement of -Recommended to be -Soil horizons are to be - Topsoil should be - Topsoil shall be -Thin layer of topsoil, - Mixing of topsoil with stockpiled growth placed on preferred replaced over the used to promote replaced to a depth with slight subsoil recommended to media after re- sites of enhanced disturbed site in the reestablishment of allowing for plant compaction to reduce produce a growth medium. contouring vegetation recovery same manner they were stored vegetation growth. - Excessive erosion placed on top stripped and stored mats if needed compaction is to be of replaced avoided. overburden. Separate practices -Yes, various -Yes, various -No. disturbance type -Yes, various -Yes, various -Yes, various -Yes, various disturbance based on disturbance disturbance types disturbance types specific practices in disturbance types disturbance types disturbance types types recorded separately. type recorded separately. recorded separately. place. recorded separately. recorded separately. recorded separately. Revegetation -Revegetate to a self- -Planting of viable -Passive revegetation - Passive revegetation - Within one year of re- -Replacement of all - Seeding recommended sustaining state stem -Active revegetation - Active revegetation contouring stored vegetation only after recontouring, -Replacement of or branch cuttings, or only used when deemed used when deemed seeding/planting should topsoil added, and soil stockpiled vegetation roots of shrubs that are necessary. necessary. occur. chemistry amended. -None added -None added - Nutrients and soil native to the area. Addition of soil -Use of fine tailings ameliorants material on sites where amendments is substances promoted, based no growth media is necessary on original site pH present -Use of woody debris. -None added - Use of pH altering 15 Chapter 3. Methodology In September 2014, a pilot study was conducted in the area east of Atlin Lake in northwestern British Columbia. During that preliminary survey 33 sites were sampled and additional sites were identified as suitable for future study. Sampled sites were placed in areas that were generally homogenous in terrain, surface materials and vegetation. Sites were selected to represent what appeared to be a wide range of times since disturbance, based on mining records, degree of overall vegetation coverage, and degree of forest floor development. These sites included variety of placer mining associated disturbances, including processed mine tailings, former sedimentation ponds, areas on which machine traffic was concentrated, and excavated sites. Consultations with the Lands and Resources Office of the Taku River Tlingit First Nation were also conducted at that time. From those discussions, it was affirmed that the recovery of wildlife habitat, and the ability of post-mining sites to support wildlife, should be considered priorities in the evaluation of ecosystem recovery. Particular interest was placed on the huntable/edible wildlife (i.e. caribou, moose, hares) that still constitute the basis for traditional foods (Trowbridge 2014). In early June 2015, field sampling recommenced, an additional 67 sites were fully sampled, including 53 which had been identified in September 2014. Sites sampled in September 2014 were resampled using updated methodology (including a search for signs of wildlife use) in the summer of 2015. Ten of the 100 sampled sites were identified as having never been disturbed by placer mining, and labelled as control sites. Control sites that were selected were in the same creek drainages as sampled 16 disturbed sites. Control sites were in the creek floodplain, on adjacent benches, or were in toe slope positions, but showing no evidence of excavation or other type of disturbance. Figure 1 shows the locations of all 100 sampled plots and associated drainage basin names. The vegetation of each site was characterized in a single using circular plot 100 m2 in area (5.64m radius), which was situated in an area of relatively uniform disturbance, substrate, and vegetation. A data collection form was refined after the pilot study had been completed, all sites sampled during the pilot study were revisited and sampled for any missing data using the updated data collection form (see Appendix I for the final updated data collection form) with additional information collected added to the existing database. Due to the large amount of variation in physical attributes between different placer mining associated disturbances it was apparent that it could not be assumed to represent a single chronosequence (i.e., with time as the only varying factor). Therefore, additional factors were recorded to assist in identifying their role in determining successional recovery and sites were categorized to different disturbance types in order to portray a series of more homogenous recovery trajectories. Multiple site and terrain characteristics were recorded at each location sampled; these attributes include the following, with methods for their determination outlined below: • Drainage basin (stream name) 17 • Geographic location, in Universal Transverse Mercator (UTM) units, m (derived from a Garmin 60s geographic positioning system (GPS) receiver) • Elevation, m (also from GPS) • Biogeoclimatic ecosystem classification (BEC) zone • Approximate soil moisture regime (SMR) and soil nutrient regime (SNR) • Slope (degrees) and aspect (degrees azimuth) • Slope position (toe, lower slope, mid, upper slope, crest and level) • Microsite relief, cm (reoccurring amplitude of elevational differences) • Disturbance type and type of reclamation performed (as categorized below) • Estimated time since disturbance, years • Soil texture (as per Canadian System of Soil Classification classes) • Summed vegetation cover, % by layer (A2, A3, B1, B2, C and D, as per BCMFRE, 2010) • “A” horizon depth, cm Multiple site and terrain characteristics were recorded at each location sampled; these attributes include the following, with methods for their determination outlined below: • Drainage basin (stream name) • Geographic location, in Universal Transverse Mercator (UTM) units, m (derived from a Garmin 60s geographic positioning system (GPS) receiver) • Elevation, m (also from GPS) • Biogeoclimatic ecosystem classification (BEC) zone • Approximate soil moisture regime (SMR) and soil nutrient regime (SNR) • Slope (degrees) and aspect (degrees azimuth) • Slope position (toe, lower slope, mid, upper slope, crest and level) • Microsite relief, cm (reoccurring amplitude of elevational differences) • Disturbance type and type of reclamation performed (as categorized below) • Estimated time since disturbance, years 18 • Soil texture (as per Canadian System of Soil Classification classes) • Summed vegetation cover, % by layer (A2, A3, B1, B2, C and D, as per BCMFRE, 2010) • “A” horizon depth, cm Figure 1. Location of all 100 sampled sites sampled east of Atlin Lake in northwestern British Columbia. All sites are associated with stream systems as named on the map, and have road access. The site classification guide by Banner et al. (1993) was utilized to determine a site’s biogeoclimatic zone, soil moisture regime (SMR) and soil nutrient regime (SNR). Determination of biogeoclimatic zone depended on elevation, UTM location relative to published map boundaries, and nearby flora. Near the centre of each plot a single soil pit was dug to a depth of 15 to 20 cm and the proportional volume of rocks and stones 19 (>7.5 cm diameter) was estimated. A 1 to 2 kg sample of the soil material representative of the 0 to 15 cm soil profile was collected in a polyethylene bag and labelled for future laboratory analysis. The growing medium retrieved was then described by hand texturing. Approximate SMR was determined along with SNR (following BCMFRE, 2010) based on slope position, coarse fragment content, soil texture, and abundance of organic matter, but minimal information from indicator plants. In order to estimate time since disturbance, annual growth rings from three of the largest trees or shrubs on or near the plots were counted. Most samples consisted of basal discs sawn off near ground level, while trees larger than about 8 to10 cm in basal diameter were sampled with a 5-mm increment corer at approximately 20 cm above the ground. Basal disks were labelled with the species and plot number using an indelible marker. Increment cores were stored in plastic straws sealed with masking tape at both ends, with labels written on the masking tape in indelible marker. A 30 m transect was laid out across plot center and along the contour of the site, keeping the elevation as uniform as possible (see Figure 2), similar to guidelines described by Anderson et al. (1979). Intersections of the transect line with erosional rills (miniature gullies in the soil surface) or areas showing evidence of sediment deposition were recorded separately, in cm of cover and then converted to a percentage of the transect showing evidence of erosion. The transect was also used to look for signs of wildlife usage to a distance of 5 m on either side of the transect, defining, in effect, a 30 m by 10 m belt transect, 300 m2. Evidence of wildlife was exhaustively searched for included animal browsing (nipped or broken-tipped stems), nests or bedding areas 20 (compressed vegetation in a circular pattern), tracks, droppings, and direct sightings, identified by species where possible. A presence or absence score was recorded for each belt transect based on the type of evidence (e.g. browsing, nesting, droppings) and species (e.g. moose, caribou, songbirds) detected. For example if evidence of moose browsing has occurred on the plot (regardless of number of browsed stems detected) it was recorded as being a moose-browsed site. Figure 2. Depiction of 30m transect laid over a sample plot (11.28m diameter). It became evident during the course of site selection that a variety of different substrates were left behind by different aspects of placer mining operations, each with a different set of environmental factors (Entrix, 1986; Farrington, 2000; Idaho Department of Lands, n.d.; Hagler Bailly Inc., 1998). The most appropriate surficial disturbance category for each plot was also recorded; sites were fit into a category based on physical site characteristics and history when applicable. Site disturbance categories include the following: 21 • Rock dumps (characterized by high percentages of cover (>50%) by large rocks, often piled); • Scraped or traffic areas (highly compacted grounds, typically old roads and heavy machine traffic areas); • Sediment ponds (high percentage of fine materials, low percentage of large rocks, often moist if not submerged/partially submerged); • Washed sand, gravel, or cobbles (high percentage of small to medium sized rock lacking fine material, often piled), • Road cuts or pit walls (ground removed to make room for a road or pit wall); • Overburden waste piles (piles of upper level ground removed to allow access to deeper materials); • Topsoil storage piles (piles of topsoil relocated to allow for access to lower materials); and • Controls (undisturbed ecosystems). Due to the small sample size for some disturbance types, some disturbance type categories were combined (e.g., sediment ponds and overburden waste piles; rock dumps and washed sand, gravel or cobbles) for some statistical analysis. An effort was made to be inclusive of sparse information existing on the vegetation of undisturbed ecosystems in the Atlin region. Specifically, efforts were made to be inclusive of data collected by deGroot and Pojar (2009) as part of sensitive ecosystem inventory conducted in the Atlin-Taku region. However, due to differences in plot size and with vegetation surveys done by more experienced botanists, quantitative comparisons to the data collected in this study were deemed unsuitable. So, for the purposes of this research, undisturbed ‘control’ conditions are based solely on the 10 undisturbed sites sampled. 22 Four subplots, each 2.0 m2 (0.8 m radius) in area were situated at each plot: one at the center of the plot, one located at the northern-most perimeter of the main plot (determined using a compass while standing at plot centre), one on the southeast plot perimeter at 120 degrees azimuth from north, and one located southwest at 240 degrees azimuth (see Figure 3). These subplots served as the reference areas for more detailed assessments of surficial cover and herbaceous layer composition. Values from those four subplots were then averaged to get a mean for that attribute at each site. The following attributes were recorded at each subplot: • Organic layer (LFH) thickness, cm • Soil penetrability (cm) • Cover of plant litter, % • Cover of dead or decaying wood, % • Cover of rock, % • Cover of exposed mineral soil, % • Cover of water, % • Cover of living vascular and non-vascular plants (% by species) Thickness of the surface organic matter layer (commonly referred to as LFH, for litter, fermentation and humus layers) was measured at several locations in each subplot, while percent cover values for each attribute including plant species were estimated by eye, calibrated by comparison among two or three researchers during the first week of each field season, and by use of the reference templates provided in the Field Manual for Describing Terrestrial Ecosystems, 2nd edition (BCMFRE, 2010). The sum count of vascular plant species encountered across the four subplots was supplemented by an exhaustive search for any additional plant species in the whole plot. As a field surrogate 23 for bulk density and compaction, a description of “soil looseness” was determined based on the depth of penetration (cm) of a common 12mm wedge-tipped tire iron, when leaned on with approximately 50 kg of weight. If a distinctive barrier was found (i.e., a large rock), a new spot was selected for the soil penetrability test. Indicators of ecosystem recovery were identified following rationale published by the Society for Ecological Restoration (SER, 2004) and Ruiz-Jaen and Aide (2005), which recognizes that multiple attributes must be assessed and that no one indicator can individually describe the health or recovery of a disturbed ecosystem. These indicators include two indicators of species composition and diversity (vascular species richness, plant community similarity compared to undisturbed reference sites), two indicators of vegetation structure (summed plant cover, structural diversity), and two indicators of ecological functionality (A-horizon depth, wildlife species richness). These six variables were selected to serve as the response variables in the statistical analysis described below, and as the primary indicators of ecosystem recovery. In order to document soil nutrient availability, PRSTM (“Plant Root Simulator”) probe pairs (Western Ag Innovations, n.d.) were installed at the three perimeter subplots at each site. These ion exchange membranes measure ion flux across separate cation-sensitive and anion-sensitive surfaces over a period of time in order to accurately integrate nutrient or metal availability (Western Ag Innovations Inc., 2013). The PRS probes were installed vertically to approximately a 15-cm depth, with the exposed membrane sampling a 2 to 12 cm depth. The PRS probes were planted in pairs (one cation probe and one anion probe) in each of the three perimeter subplots. Three pairs of PRS probes (six individual probes) were installed at each site, with soil 24 backfilled by hand to ensure good membrane contact and then compressed by hand after installation. Figure 3. Depiction of subplots associated with a main plot. Four subplots were utilized one at the center of the plot, one located at the northern-most perimeter of the main plot, one southeast at 120 degrees’ azimuth from north, and one located southwest at 240 degrees’ azimuth to north. PRS probes were planted between June 15th, 2015 and June 23rd, 2015; PRS probes (600 individual probes) were later retrieved between July 20th 2015 and July 23rd, 2015. Once retrieved, PRS probes were washed with de-ionized water, stored in individual polyethylene bags, labelled and refrigerated prior to transport. PRS probes were packaged and transported on ice, and then shipped to Western Ag Innovations (Saskatoon, Saskatchewan) for processing. 25 Laboratory methods Soil samples, basal disks and increment cores were stored in lightproof plastic bins in the back of a canopied pickup truck, until logistics could allow for delivery to the UNBC Enhanced Forestry Laboratory (Prince George, British Columbia), approximately one to two months after collection. Soil samples were allowed to air dry for 24 hours at room temperature before being labeled and stored in polyethylene bags prior to transport to the UNBC Enhanced Forestry Laboratory. Basal disks were stored in breathable containers to reduce the chance of mold. Similarly increment cores stored in plastic soda straws with slits cut into them to allow for aeration to reduce the chance of mold occurring. At the UNBC laboratory, basal discs were sanded using a progression of sandpaper grits (80, 120, 240, 300, and 400 in succession), whereas increment cores were initially mounted into slotted wood holders by means of wood glue, then sanded. Sanded samples were viewed under a dissecting microscope to count annual growth rings. The maximum number of latewood rings perceivable across three separate radii of basal discs was considered to be the true age of that tree or shrub. For increment core samples, an extra 2 years was added to the ring count to estimate age for willows (Salix sp.), poplars (Populus spp.), and pine (Pinus contorta Douglas ex Louden), while 3 years was added for spruce (Picea spp.) and fir (Abies lasiocarpa (Hook.) Nutt.), to adjust for the time needed for seedlings to grow to sampling height of 20 cm (McCarthy et al. 1991). Recognizing that there is typically a delay between site closure after active mining and the establishment of early seral vegetation, an effort was made to determine 26 the true date of last mining or mine reclamation activities on the sites sampled (Hardy Associates, 1978; Entrix, 1986). Four plots were able to be aged using available sources of historical information or local knowledge (see Table 2); from this information, it was determined that a mean time for woody species (trees or shrubs) to establish was 5 years after disturbance (Table 2). Table 2. Estimation of average offset between tree or shrub ages and evidence of the last mining or reclamation activity. Plot ID 15-06 Year Oldest Tree or Shrub Established 2004 Year of Confirmed Last Industrial Activity 2000 22-02 1994 1989 22-04 22-05 1996 1997 1991 1991 Source of Mining Completion Confirmation Sector Resources Canada (Murray Straughn) Min. of Energy, Mines & Petroleum Resources Annual Report; confirmed by Doug Flynn Clive Aspinall Clive Aspinall Lag Time to Tree or Shrub Establishment 4 years 5 years 5 Years 6 Years Based on this limited calibration, the estimated time since disturbance assigned to each site, except the four known sites, is the maximum age derived from basal disks or increment cores plus 5 years to account for the delay in woody plant establishment. Retrieved soil samples were allowed to air dry in the Enhanced Forestry Laboratory at the University of Northern British Columbia prior to being passed through a 2 mm sieve, then subsequently a 0.05 mm sieve. The weight of each material was recorded in three categories (larger than 2 mm, 0.05 mm to 2 mm, and smaller than 0.05 mm) and expressed as a percentage of the total sample mass (Gee & Or, 2002). A portion of material that had passed through the 2 mm sieve was used in further tests, 27 including pH, electrical conductivity, and organic matter content. A 10 g portion of material that had passed though the 2 mm sieve was combined with 20 ml (2:1 ratio) of distilled water (pH 5.4), creating a suspended solution. The solution was allowed to settle then subsequently tested for pH and electrical conductivity using a Hannah Instruments HI98130 pH/Conductivity meter. In order to determine a soil sample’s organic matter content a portion of the material that had passed through the 2 mm sieve was then combusted to determine mass loss on ignition (LOI), following methods outlined by Ball (1964). Subsamples weighing approximately 90 to 100 g were first dried overnight in a drying oven set at 105 °C to remove any remaining moisture before weighing. These subsamples were then weighed on a top loading draft-free scale to the nearest 0.0001 g. Once weighed, subsamples were placed in a muffle furnace set to 360 °C for 2 hours. Subsamples were then allowed to cool; once their temperature had fallen below 150 °C they were once again weighed. The change in mass was then expressed as a percentage of the oven-dry mass, for which the difference is assumed to represent the mass of all organic matter which had been lost by way of ignition in each sample. Plant root simulator (PRS) probes were packed with frozen cold packs and sent to Western Ag Innovations Inc for processing. Western Ag Innovations processed the PRS probes by elution for one hour with 0.5 N HCl / 2 M KCl, followed by automated colorimetric analysis for ammonium and nitrate, and inductively-coupled plasma spectrophotometry, atomic absorption spectrometry, and flame emission spectrometry for other ions. Results are reported in µg ion/10 cm2/burial period for individual anions (NO3-, H2PO4-, SO42-) and for individual cations (NH4+, K+, Ca2+, Mg2+, Al3+, Fe3+, 28 Mn2+, Cu2+, Zn2+, B(OH4)3+, Pb2+, and Cd2+) (Western Ag Innovations Inc., 2013). Nutrients are referred to using only the element of ecological interest, rather than the ionic form (e.g., P denotes H2PO4-, B denotes B(OH4)3+), except where necessary in order to distinguish different forms (i.e., NH4 and NO3). Ammonium and nitrate values were summed to denote total available N. Although the length of time probes remained in the ground ranged from 29 to 38 days (mean 33.3 days), no correction for differences in the length of sampling period was considered necessary or possible, as ion uptake is not linear over time, and is generally more sensitive to periods of wetting and drying than time per se (pers. comm., K.S. Arachchilage, R&D Coordinator, Western Ag Innovations Inc., 11 September, 2015). Statistical methods In order to prepare the data (Appendix II) for later statistical analyses, several aggregate statistics and indices were calculated. Data collected from the subplots were averaged and the singular averaged value was used to calculate later statistics and indices. The mean and 95% confidence limits for all 6 indicators of ecosystem recovery based on the 10 control sites were calculated, to be used as a reference to ‘target’ ecosystem conditions. These calculated values were later used to interpret statistical projections of the amount of time needed for ecosystem recovery to reach a target state. Structural diversity of vegetation layers and ground cover types was quantified using the Shannon H’ index of diversity (Equation 1) for each site (Magurran, 1988). Calculation of the H’ index used the percent cover of the tree layer, shrub layer, herb layer, cryptogam layer, dead wood, rock cover, and water cover as input “layers”: 29 H’ = -∑(piln[pi]) (1) where pi is the proportion of total cover provided by layer i, and ln is the natural logarithm function (log base e, 2.71828). A mean similarity metric was calculated utilizing the distance matrix option (McCune et al., 2002) in PC-ORD Version 6 (McCune & Mefford, 2011) in order to compare plant community similarity between disturbed sites and the undisturbed control sites. A similarity matrix utilizing a relativized Sorenson index comprised of all combinations of individual plot comparisons was constructed then averaged for individual plot comparisons to the 10 sampled control sites. Further data exploration was then conducted using a rank correlation analysis, through the determination of Spearman’s rho using the R Statistical Platform (R Core Team, 2015) in order to examine the relationship between various continuous site variables and ecosystem recovery indicators. Data that were recorded as categorical in nature were also explored utilizing a one-way analysis of variance (ANOVA), followed by post-hoc Tukey multiple comparison tests utilizing the R Statistical Platform (R Core Team, 2015). One purpose of statistical analysis was to determine the amount of time in years needed to achieve various thresholds of ecosystem recovery, under differing circumstances. This method focused on the above indicators of ecosystem recovery and their response to estimated time since disturbance. This method was completed using interpolation or extrapolation of least squares linear regression analysis using the R Statistical Platform (R Core Team, 2015). In exploring the trends in this survey-based 30 sample data, all relationships with p<0.05 are considered significant, and those with p<0.10 are considered worthy of note and further investigation. In an effort to determine which site characteristics were most important to ecosystem recovery after placer mining has ceased, a classification and regression tree (CART) method was undertaken utilizing the rpart module (Therneau et al., 2015) in the R statistical framework (R Core Team, 2015). Using the available suite of potentially predictive site attributes, the CART trees sequentially partition the data into increasingly homogeneous subsets of a given response variable (De’ath & Fabricius, 2000), in this case the individual indicators of ecosystem recovery. The goodness of fit (proportion of the total variance explained) for each CART tree is described as an R 2 statistic (Steinberg, n.d). The root mean square error (RMSE) statistics is used to characterize the variability in each terminal subset of the data. As this is a “data mining” approach to statistical analysis and involves no inferential hypothesis testing, there is no calculation of the probability of Type I error, or p value (Banerjee, 2009). The factors identified by the CART approach are those which distinguish various subsets of high and low values for a response variable in various subsets of the data. In this manner, the method also facilitates the identification of thresholds that help identify conditions that promote or detract from ecosystem recovery. 31 Chapter 4. Results Generally speaking, the undisturbed control sites were climax ecosystems, often dominated by a few late seral species such as Picea glauca (Moench) Voss, Pinus contorta Douglas ex Louden, Linnaea borealis L. and Vaccinium vitis-idaea L. (see Appendix III for further details). Control sites also supported high levels of vegetation and plant litter cover with minimal levels of bare mineral soil exposed, along with welldeveloped soil horizons. Mean conditions on undisturbed sites were calculated for each indicator of ecosystem recovery. For vascular plant species richness, a mean of 10.2 species across a 100 m2 plot, a range of 5 to 18 (Table 3) was determined on control sites. Mean compositional similarity to all other control plots averaged 0.235 (Table 3), with a range of 0.146 to 0.317. Summed vegetation cover had calculated mean of 173% (Table 3) and a total range of 95 to 255 on undisturbed sites. Structural diversity was found to have a mean of H’ index of 1.458 and ranged from H’ indices of 1.095 to 1.807 (Table 3). A horizons on undisturbed sites were typically well developed and had a mean depth of 10.2 cm (Table 3), with a range of 5.0 to 14.8 cm. Table 3. Representative conditions on undisturbed sites in Atlin area (n=10). Indicator Mean Vascular plants richness, spp. count per 100 m2 Compositional similarity to controls Summed vegetation cover, % Structural diversity, H’ A horizon depth, cm Wildlife species richness, spp. sign count per 300 m2 S.EM. 95% C.L. 3.2 Lower 95% C.L. 7.0 Upper 95% C.L. 13.4 10.2 1.4 0.235 0.017 0.038 0.197 0.273 173.31 1.458 10.2 0.977 14.5 0.079 1.2 0.083 32.8 0.179 2.7 0.166 140.5 1.280 7.5 -0.811 206.1 1.637 12.9 1.143 32 Disturbed conditions There was an immense amount of variability among the 90 identified sites disturbed by placer mining. Sites varied in terms of topographical characteristics, amount of vegetation cover (6% to 160%), and in time since disturbance (9 to 76 years). Figure 4 shows a site estimated to have been abandoned about 10 years previously, in which the growing medium consists of mounded washed or sorted gravels. The site in Figure 4 supports a sparse vegetation cover with bare mineral substrate exposed in many regions. In contrast, Figure 5 portrays a site several decades after disturbance that contains established shrub and tree layers. The site in Figure 5 is also more gently rolling in terms of microsite relief in comparison to Figure 4, and has significantly less bare mineral soil exposed. Figure 4. Plot 15-01 demonstrates a site with sparse vegetation and exposed bare mineral soil in many area, the estimated time since disturbance is ten years. 33 Figure 5. Plot 15-10 demonstrates a plot with an established tree and shrub layer and minimal bare mineral soil exposure. The estimated time since disturbance is 45 years. Similarly, there was a great amount of variation among site substrates apparently related to disturbance type. Out of the 90 disturbed sites sampled, 45 were identified as rock dumps (piles of large cobbles or boulders), 22 as scraped or traffic areas (machine traffic areas or roads), 7 as settling ponds (sedimentation ponds, typically with high levels of fine materials), 7 as overburden/waste piles (piles of topsoil or glacial till removed to access lower placers), 6 as washed sand/gravel/cobbles (piles consisting of mine spoils from the washing process), and 3 as road cuts or pit walls (from which material had been cut away to make way for a road or pit). After disturbance, the majority of sites sampled were left for natural processes to complete ecosystem recovery (n=72). Out of the 18 sites that showed evidence of reclamation efforts, 13 showed signs of contouring or re-sloping, and 5 34 showed evidence of soil or overburden replacement. Wildlife activity as evidenced by observation of sign of different species or species groups had a mean of 0.99 species per site, with a range of 0 to 3, and with 60% of the undisturbed sites and 61% of the disturbed sites showing evidence of moose browsing (Table 4). Table 4. Compilation of wildlife activity evidence presence found on sites, for both disturbed and undisturbed sites. Evidence of wildlife activity Disturbed sites (n=90) Moose 55 Grouse 12 Songbird 2 Grouse 1 Bear 5 Hare 5 Ground squirrel 1 Waterfowl 2 Caribou 4 Wolf 1 Porcupine 1 Control sites (n=10) 6 0 0 0 0 4 1 0 0 0 0 Regression analysis A general trend of increasing summed vegetation cover was detected as time since disturbance (TSD) increases (see Figure 6), although the relationship between the two is marginally significant and very weak (p=0.0782, R2= 0.0348, n=90). The pronounced variation in the amount of summed vegetation cover among multiple sites of similar TSDs suggests that there are more factors than just time affecting ecosystem recovery rates. 35 Figure 6. Summed plant cover (sum of tree, shrub, herb and moss/lichen layers) as related to time since disturbance and categorized by disturbance type on placer mined sites east of Atlin Lake, BC. Other indicators of ecosystem recovery showed more distinct trends with TSD, specifically structural diversity (p<0.0001, R2=0.2542, y=0.01267x+0.67402), and compositional similarity to control sites (p=0.0018, R2=0.1056, y=0.00131x+0.05939). Extrapolations of significant regression equations to the means and confidence limits for undisturbed (control) sites are presented in Table 5. These forecasts include the intersection of least-squares regression lines with mean control levels, upper 95% confidence limits intersections with lower 95% confidence limits for control values, and regression line intersections with 70% of mean control levels (a target designating successful recovery according to the placer mining reclamation standards of Idaho Department of Lands, n.d.). Regression-based projections show that all indicators of ecosystem recovery develop on separate timelines (Table 5). 36 Figure 7. A-horizon depth as related to estimated time since disturbance after placer mining. Abbreviations are as follows: SP is settling ponds, OBWP is overburden waste piles, RD is rock dumps, WSGC is washed sand/gravel/cobbles, STA is scraped traffic area and RCPW is Roadcut and pit walls. Accounting for all utilized indicators of ecosystem recovery, a mean estimate of 101 years was derived as the length of time needed for disturbed sites to recover from post-mining conditions to mean control site conditions. However, when analyzed separately the timelines for each individual indicator of ecosystem recovery range widely. As an example, an estimated 62 years is needed for the indicator structural diversity to reach mean control conditions, 106 years for A-horizon depth (see Figure 7) and potentially as high as 134 years for compositional similarity to controls. 37 Table 5. Significant regressions of indicators on estimated time since disturbance, and projected recovery times to alternative thresholds. Recovery Indicator structural diversity, H’ compositional similarity to controls A horizon depth, cm Average time to recovery: structural diversity, H’ structural diversity, H’ A horizon depth, cm A horizon depth, cm Average time to recovery: Linear Regression Analysis R2 Intercept Slope Time for Regression to Achieve Control 95% 70% Ctl. Means LCL Means 61.9 47.8 27.4 134.2 105.2 80.3 Time for Upper C.L. to Achieve Control 95% 70% Ctl. Means LCL Means 38.3 27.9 12.9 72.8 54.9 39.5 Subset P All All <0.0001 0.0018 0.2542 0.1056 0.67402 0.05939 0.01267 0.00131 All All 0.0185 0.0615 -0.19522 0.09816 105.9 100.7 78.8 77.3 74.7 60.8 45.8 52.3 30.9 37.9 28.7 27.0 0.29585 0.63481 0.78297 -2.30453 0.03487 0.01322 0.08277 0.1876 33.3 62.3 132.7 66.7 73.8 28.2 48.8 100.5 52.5 57.5 20.8 29.2 95.7 50.3 49.0 15.4 34.4 55.7 34.4 35.0 11.9 25.0 38.8 25.3 25.3 6.9 11.4 36.3 23.9 19.6 Disturbance Type SP & OBWP 0.014 0.5876 RD & WSGC <0.0001 0.2962 RD & WSGC 0.03331 0.0893 STA 0.0016 0.03586 Disturbance Type Subgroups Reclamation treatment structural diversity, H’ NAT <0.0001 0.276 0.68317 0.01248 62.1 47.8 27.1 37.1 26.8 11.8 structural diversity, H’ PB 0.0352 0.8173 -0.78361 0.07366 30.4 28.0 24.5 18.8 16.9 14.1 A horizon depth, cm NAT 0.0007 0.1529 -0.93628 0.09275 120.1 91.4 87.1 77.0 56.9 53.9 Average time to Reclamation treatment subgroups 70.9 55.7 46.2 44.3 33.5 26.6 recovery: Disturbance Types: SP = Settling ponds, OBWP = Overburden waste piles, RD = rock dumps, WSGC = washed sand/gravel/cobbles, STA = scraped traffic areas, road cuts and pit walls. Reclamation Treatments: NAT = left for natural recovery, PB = surface materials pulled back 38 Classification and regression tree (CART) analysis results Variables, categorical distinctions, and thresholds deemed crucial by the CART analysis emerged with data segmented into homogenous categories for each indicator of ecosystem recovery. Vascular plant richness seemingly depends on one of the most multifaceted sets of conditions for explanation, with nine categories designating mean values ranging from 5.7 to 13.4 species per 100 m2 plot. Notably, microsite relief defines the first break point, with sites having greater than 75 cm of re-occurring relief patterns supporting an average of 13.4 plant species per 100 m2 plot. Other important variables include elevation, slope, aspect, and soil looseness; for example, sites at elevations >915m and on slopes >5o or with compacted soil (penetrability <2.5 cm) supported averages of only 5.7 to 9.2 plant species across four subplots (Figure 8). For compositional similarity to controls, seven categories were identified, with mean similarity to controls ranging from 5% to 15%. The first break point and important factor is denoted by disturbance type (rock dumps in particular), with the second most important factors selected being elevation and TSD, then lastly microsite relief (Figure IV-1 in Appendix IV). The CART analysis helps illustrate how compositional similarities improve as time since disturbance increases, with three notable breaks at 16 years, 22 years, and 27 years. As a visual representation, Figure 9 shows a CART tree that uses five factors to distinguish seven different subgroups of the recovery indicator of summed vegetation cover. Based on this analysis, means were found to range from 23% summed vegetation 39 cover (on southwest-facing sites on mid- or higher slope positions with non-loam soils after more than 30 years since disturbance) to 93% cover (for level, toe-slope or depression sites on loamy soils). In order of importance, the factors deemed crucial by CART analysis to summed vegetation cover are soil texture > TSD and slope position > aspect and microsite relief. The CART analysis for structural diversity derived seven categories in the end, ranging from H’ indices of 0.57 to 1.40 (see Figure IV-2 in Appendix IV). Notably the first break point is time since disturbance, with a derived threshold of 23 years since disturbance. Another interesting threshold relates to elevation, with sites under 927m in elevation having the highest values. It is also worth noting that the disturbance type washed sand/gravel/cobbles (typically devoid of fine materials) appears as a break point, with that specific disturbance type earning the lowest mean structural diversity values. In order of importance the factors deemed crucial for structural diversity are: TSD > disturbance type and elevation > slope position and soil looseness > soil texture. 40 Figure 8. Classification and regression tree for factors influencing vascular plant richness (species count across 100 m2 plot) on post-mining sites east of Atlin Lake, B.C. R2 = 0.39. A-horizon development was shown to respond to a few different factors (Figure IV-3 in Appendix IV). In order of importance, those factors are microsite relief > time since disturbance and slope position > reclamation efforts. Five categories were determined ranging from mean A-horizon depths of 0.2 cm to 9.3 cm. It is worth noting that sites situated at a level, lower or toe slope position have the highest mean Ahorizon depth of 9.3 cm. 41 Figure 9. Classification and regression tree for factors influencing summed vegetation cover (%) on post-mining sites east of Atlin Lake, B.C. R2 = 0.41. There was lower wildlife species richness noted at sites with an elevation under 888 m, and higher wildlife richness at elevation greater than or equal to 1131 m. Six terminal categories with mean wildlife richness scores ranging from 0.444 to 1.8 were identified (Figure IV-4 in Appendix IV). With disturbance type identified as the first important factor, in order of importance the remaining important factors are elevation > aspect > soil texture. A few influential factors were found to be reoccurring within the CART analysis as being deemed crucial to the selected indicators of ecosystem recovery. To synthesize these results, the appearance of each factor along with its associated importance level was compiled with a weighted mean importance rank calculated for each factor (Table 6). In order of importance, the reoccurring factors are time since disturbance > microsite relief > elevation > slope position > soil texture > aspect > disturbance type > reclamation done > soil looseness > slope steepness. 42 Table 6. Compilation of ranked importance of attributes on six indicators of post-mining ecosystem recovery, based on their position in CART trees (Figures 8 and 9 and Appendix IV). Attribute Compositional similarity to controls Vascular species richness Elevation Slope Slope position Aspect Microsite relief Soil looseness Soil texture Disturbance type Reclamation done 2 2,4,5 3,6 TSD* Number of terminal subsets 2,3,3 7 CART goodness of fit, R2 0.46 4 5 1 4 Summed vegetation cover Structural diversity A-Horizon depth 2 2,3 4 4 3 1 4 2 1 Wildlife species richness Mean rank Importance ** Weighted mean rank importance *** 2,3 5.19 7.75 5.92 5.17 5.17 6.50 6.83 5.33 7.50 2.38 3.60 2.49 2.98 1.98 3.55 2.65 3.16 3.38 3.61 1.90 2 1 1 1 1 3 9 2 7 1 7 2 5 6 0.39 0.41 0.51 0.30 0.63 * TSD = Time since disturbance **with low values denoting higher priority in the CART construction process, and attribute absence assigned a nominal weight of 7 (reflecting the fact none of the CART trees had more than 6 levels of branching). *** rank values weighted by the CART model’s unexplained variance (1.0-R2), so lower values (higher rank) denote a combination of greater importance and greater confidence in the results. 43 Chapter 5. Discussion There was a large amount of variation found among the 90 disturbed sites sampled. Each of the 90 disturbed sites is in the process of recovering under a unique set of environmental conditions, each with its own mining history, terrain characteristics, soil characteristics and plant communities. It is clear that the sampled sites can not be assumed to represent a single chronosequence with time as the only varying factor. The categorization of sites into disturbance type based on similar physical characteristics allowed for the assumptions of a chronosequence to be met for each disturbance type, allowing for recovery trajectories for separate disturbance types to be created. The selected indicators of ecosystem recovery appeared to be responding to different variables and are developing at differing rates. For example, structural diversity, which denotes multiple substrates and colonization by multiple plant growth forms is one of the quickest to recover (30 to 62 years) (Table 5), while A-horizon development (which is indicative of nutrient cycling and organic matter accumulation and decomposition) needs a longer period of time to recover (106 to 133 years depending on disturbance type and available substrate). Similarly, compositional similarity of disturbed sites to undisturbed sites requires an estimated 134 years (Table 5) to recover to mean undisturbed conditions. Overall the average time needed for disturbed sites to recover from post-mining conditions to mean control site conditions is 101 years; estimates can be further refined when accounting for different disturbance types or different substrates. Doing so brings the mean time needed for disturbed sites 44 to recover to mean control conditions down to 71 to 74 years. The estimated average time projected for structural diversity and A-horizon recovery to mean control levels is 91 years (Table 5), in perfect agreement with the 80 to 100-year estimate by Entrix (1986) for disturbed placer mining sites in Alaska to recover into a climax ecosystem without the use of reclamation efforts. Comparing the recovery trajectory for sites with differing substrates, distinctions can be made between substrates containing fine materials (settling ponds and overburden waste piles, n=14), where a mean of 33 years is needed for structural diversity to recovery to target conditions, compared to substrates lacking fine materials (primarily rock dumps and washed sands, gravels, cobbles, n=51) where a mean of 62 years is needed for structural diversity to reach target conditions. Distinctions can also be made when comparing sites originating from different disturbance types. For the Ahorizon recovery indicator, rock dumps and washed sands, gravels and cobbles take an estimated average of 133 years (see Table 5) to meet target conditions, while scraped traffic areas or road cuts and pit walls are expected to require an average of 67 years. Due to the low sample size, it is more challenging to gauge the effectiveness of reclamation treatments (n=18, 13 re-contoured sites and 5 with soil or overburden pulled back) compared to natural processes. Nevertheless, it is forecast that sites with soil pulled back or overburden replaced would take an estimated 30 years to recover to mean control levels, in contrast with the 62 years forecast for naturally recovering sites. Each indicator of ecosystem recovery seemed to be influenced by factors previously recognized by numerous authors and included in the literature review (e.g., 45 Entrix, 1986; Densmore, 1994; Idaho Department of Lands, n.d.). Many of these factors are also reflected in published best management practices for many jurisdictions globally (e.g., Farrington, 2000; SMEGAC, 2012; BCMFLNOEM, 2014). However, the utilization of several different indicators of ecosystem recovery has allowed for separate factor identification based on which indicator of ecosystem recovery is considered important, along with different projections of estimated recovery time to target conditions. Utilizing the identified factors with their associated thresholds has the potential to provide guidance on how post-mining sites might be categorized or manipulated in order to enhance ecosystem recovery after a placer mining disturbance. Vascular plant species richness was sensitive to variations in elevation, and was generally found to be higher in the BWBS zone than in the SWB zone (Appendix V). Soils with a moderate amount of looseness, and sites with a generally SW-facing aspect were also found to have high species counts. In the northern hemisphere, south-facing slopes receive more direct sunlight and thus have higher temperatures, higher rates of evapotranspiration, and thus retain lower amounts of soil moisture compared to northfacing slopes (Måren et al., 2015). Surprisingly, species richness was also found to be greater on sites with high amount of topographical micro-relief (particularly >75 cm). Sites containing high amounts of reoccurring topographical micro-relief, in other words sites with a rough, undulating surface, offer plant seeds/propagules multiple benefits compared to a smooth homogenous surface during the early stages of plant establishmentRough surfaces offer more nooks and crannies for seeds to get caught in and begin the germination process compared to a smooth homogenous surface (Harper et al., 1965). Similarly, high amounts of micro-relief may also aid in the capture and 46 retention of organic litter (Hamrick & Lee, 1987), and the capture and retention of snow during winter months. Due to the variability within a site with high amounts of microrelief, multiple micro-climates can be formed, with areas of lower or higher evaporation (due to factors such as reduced exposure to direct sunlight). The created microclimates allow for higher species richness as plants that prefer xeric, mesic, or hydric environments can be established in close proximity at the same site (Dowling et al., 1971). Most disturbed sites have low compositional similarity to the undisturbed control sites; however, this relationship appears to be very sensitive to time since disturbance (with divisions at 16, 22 and 27 years since disturbance). Oddly, microsite relief and elevation appear to have contrasting positive and negative effects (based on the sign of rank correlations presented in Table VI (see Appendix VI) when considered for all subsets of the data. However, both emerge as positive effects when considered for non-rock dump disturbance types and within the time since disturbance range of 22 to 27 years (Figure IV-1 in Appendix IV). In the case of summed vegetation cover as a key indicator, higher values of vegetation cover were generally found on sites with increasing time since disturbance, in particular sites that are greater than 30 years old (Figure 9). Higher values of summed vegetation cover were also found on sites with minimal reoccurring microsite relief (particularly less than 28 cm), sites situated on a lower or toe slope position, and sites containing a loose loamy soil. In contrast, summed vegetation cover was found to be lower when situated on a crest, upper, or middle slope position (Figure 9), with soils 47 consisting of high coarse soil (>2 mm) content (lacking fine materials), and with soils demonstrating a high pH level. Lower amounts of summed vegetation coverage on sites containing mainly coarse soil (>2 mm) comes as no surprise, as substrates that are chiefly comprised of coarse material support large interstitial spaces which constrains many natural processes, such as plant growth and nutrient cycling (Oades, 1984). The Shannon H’ index (which is based on the cover of living vegetation layers, plant litter, and dead wood on the ground) was utilized to measure the structural diversity of sampled sites. Higher H’ indices values were found on older sites, particularly sites ≥23 years since disturbance, sites at a low elevation (particularly ≤927 m), on overburden waste pile disturbance types, and on sites situated on level, lower or toe slope positions (see Appendix V). Lower H’ indices were associated with younger sites, sites situated on crest, upper or mid-slope positions, and with disturbance type falling into the washed sand, gravel, or cobbles category. Soil development trends were variable over space and with time since disturbance. The interpretation of such trends is confounded by the fact that differing types and amounts of materials were apparently spread in the course of exploration and mining activities (Trowbridge, 2014). The majority of disturbed sites (n= 72) showed few or no indications of soil development with a definable A-horizon (e.g., Figure 10). Interestingly, A-horizon depth was the only indicator to deem reclamation efforts as a crucial factor, this likely representative of sites where topsoil was replaced, rather than in situ topsoil development. 48 Nevertheless, 18 disturbed sites showed evidence of an A-horizon (e.g., Figure 11), and 19 of the sites had a measurable organic horizon (LFH). A-horizon depth showed a strong relationship with reclamation done, and was found also to be deeper on sites with small amounts of microsite relief, specifically <7.5 cm (probably reflecting the re-contouring operations conducted), or if situated on a level, lower, or toe slope position. In particular sites capped with overburden material seem to have well “developed” soil horizons, while sites consisting of large cobbles/boulders or washed sand/gravels/cobbles show little to no signs of soil development. Studies focused on reclaiming sites devoid of fine materials further showcase the importance of soil reconstruction, either by means of a topsoil cap or replacement of the originally removed organic layer. Without active reclamation efforts that involve the re-spreading of removed soil and organic materials is it estimated to take 120 years for topsoil with a depth of 10 cm to develop (Table 5), similarly 87 years is needed for a depth of 7 cm and 57 years is needed to be within the range of natural variability of control sites. The placement of a topsoil cap or reconstruction of soil horizons were found to offer many ecological benefits elsewhere, including improved water retention, reduced soil erosion, accelerated plant establishment and plant growth (Johnson, 1987; Christensen et al., 2013; Cooper & Van Haveren, 1994). Furthermore, it was found that vegetation establishment and growth was most vigorous when both topsoil and subsoil horizons have been reconstructed (Halvorson et al., 1986) 49 Figure 10. Soil profile in Plot 15-05, showing negligible soil development; estimated time since disturbance is 20 years. Figure 11. Soil profile in Plot 20-08, showing an apparently well-developed A horizon (to 9 cm), and B horizon (to 20 cm); estimated time since disturbance is 33 years. 50 Wildlife species richness is evidently greater at higher elevations, particularly ≥1131 m (see Figure IV-4 in Appendix IV). Wildlife activity overall was also found to be more prevalent in the more eastern creek drainages (see Appendix VI). Though these trends reflect distance from human activities, other trends are more challenging to interpret: E.g., higher levels of wildlife activity were found on sites with low levels of soil looseness. A greater abundance of wildlife sign on slopes with a southwestern aspect may reflect a preference of birds and mammals of such sites for sunning (Kennedy, 1969). It is worth noting that wildlife activity seems to occur under most post-mining conditions, however much of it may be transiting, making it difficult to infer any effects with regard to ecosystem recovery. While overall wildlife habitat suitability is more of a stand or landscape attribute than can be assessed on a transect, the presence of wildlife sign on a site at least indicates that conditions there are not adverse to native animals. A primary reason for this is the large innate home range size of the region’s fauna compared to the relatively small area and narrow span of mining-disturbed habitat and the very small area for which site attributes were characterized. For example, the average annual home range size for a single moose can vary from 2 to 20 km2 (Cederlund & Sand, 1994), with variables such as size and body weight also affecting home range size (Harestad & Bunnell, 1979). With wildlife sign sampled from areas of only 300m2, and each species of animal having individual habitat requirements, the collected data only allows for insight into a very small fraction of a species’ annual habitat use. 51 Defining a target for ecosystem recovery is important to measuring recovery as it dictates at which thresholds the indicators of ecosystem recovery are considered recovered. Utilizing a target such as 70% of the control means may be deemed as an acceptable target for some ecosystem recovery scenarios (Idaho Department of Lands, n.d.). Alternative targets may also be selected such as recovery to a disturbed location’s range of natural variability (Landres et al., 1999) (overlapping of 95% confidence limits), which encompasses a site’s past conditions and processes. Another alternative endpoint for ecosystem recovery includes recovery to a self-sustaining ecosystem, which can be much more difficult to determine (Singh et al., 2016). Differences in the defining the level at which disturbed ecosystems are deemed recovered can seriously reduce or extend the time needed for acceptable recovery. Moving forward, social decisions will need to be made in regard to which indicator of ecosystem recovery to strive for, and what level of recovery is acceptable to deem a disturbed site recovered. This research utilized control site means and their associated upper and lower 95% confidence limits, as well as 70% of the control means as alternative endpoints for projected ecosystem recovery. Other endpoints for ecosystem recovery may be selected based on the desired degree of reclamation, such as to a self-sustaining ecosystem (Singh et al., 2016), reducing or extending the time needed to deem a site as recovered. In addition to deciding on how rigorously ecosystem recovery should be defined, another decision also needs to be made with regard to which ecosystem system recovery indicator (or combination of indicators) should be utilized or emphasized. As the ecosystem indicators progress at different 52 rates and embody different ecosystem conditions and services, the decision needs to be made as to which indicator or indicator combination best suits local goals. 53 Chapter 6. Conclusions and recommendations Each indicator of ecosystem development has a distinctive set of factors that are associated with ecological recovery after placer mining. Summarizing the factors corresponding with high values of selected indictors has allowed for the determination of which factors are most pertinent to ecosystem recovery in general, and in some cases those which are detrimental to ecosystem recovery. Optimizing the factors so identified can reduce the amount of time needed for a disturbed site to reach mean undisturbed conditions, or any other defined management target. For example, structural diversity is estimated to take a mean of 62 years (see Table 5) to reach mean control levels on washed sand/gravel/cobbles. However, on sites where fine material is present, the extrapolated time needed to achieve those same mean control conditions is only an estimated 33 years. Some individual factors can be manipulated towards an optimal state in order to promote ecosystem recovery. Factors that can be manipulated include slope, soil looseness, microsite relief, presence of organic material and the presence of fine materials. However, there are also factors affecting ecosystem recovery that can not be manipulated; these include factors such as elevation, time since disturbance, aspect and geographic location. The recommendations suggested by this research stem from trends detected within the collected data. Many of these identified trends agree with the trends outlined in the literature. Utilizing these trends, the recommendations for ecosystem recovery after placer mining disturbances can be reduced to a few suggestions, which amplify 54 and give local relevance to the best management practices identified for reclamation after placer mining everywhere (Table 1). Based on the superior performance of sites capped with re-spread materials any and all soil, organic materials, and removed overburden should be stockpiled and stored for later re-spreading on mine spoils, if at all possible. Storage times should be kept to a minimum in order to protect the viability of rhizosphere bacteria, mycorrhizal spores, and plant seeds present in the removed material. Care should also be taken to ensure that stockpiled materials are not piled more than 1 m high in order to reduce the likelihood of anaerobic conditions and further protect its biotic viability (Thurber Consultants et al. 1990). This research was not in the position to test the optimal length of time or optimal storage depth for stockpiled materials. If the site prior to disturbance contains any plants known to resprout vegetatively, such as willows or sedges, that material should also be stored under light soil cover and saved for later re-spreading. If kept alive, much of this material could re-establish and aid in reducing the time needed to reach a recovered state from a disturbed state. Materials that are stockpiled for later use should be used to cap sites where topsoil/overburden was removed. Settling ponds constructed to treat the water used in gold processing are important not only as a method of keeping sediments from flowing downstream through creek waters, but they also serve as a way to capture fine materials that can be used later for reclamation efforts, particularly on sites currently devoid of fine materials. 55 When re-contouring a disturbed site or waste piles (especially rock dumps and washed sands/gravels/cobbles), care should be taken to ensure that gentle slopes within the range of 6 to 11 degrees or less are created. In an effort to protect against soil erosion, further care should be taken to ensure that machine traffic lines run parallel to the contours, not up and down the slopes. No matter the reclamation effort undertaken (re-contouring slopes, spreading of topsoil cap, or ripping of compacted traffic areas), care should be taken to ensure that the surface is left fairly rough and loose, with a hummocky topography containing reoccuring microsite relief 25 to 30 cm in amplitude. Based on the strong divergence of projected times to reach mean undisturbed conditions on sites with and without fine materials (Table 5), sites that are typically devoid of fine material (such as the rock dump and washed sand/gravel/cobble disturbance type) should be prioritized for reclamation. Despite the acknowledgement of identified general trends, due to the high variability observed in sampling the field sites encountered east of Atlin Lake, ongoing monitoring and regular inspections are suggested in order to identify areas remaining barren of vegetation. Further remediation treatment may be warranted on a spot by spot basis as deemed necessary. Suggestions for future research Low sample size in subsets of the data limited the explanatory power of the statistical methods undertaken. In particular, the relatively low number of sites where reclamation activities had returned soil, overburden or organic matter back to the site 56 (n=5) limited the degree to which statistical methods could evaluate the effectiveness of these treatments. If recommendations described above are implemented and if monitoring is undertaken as suggested, then a growing body of data will allow improved evaluation of the effectiveness of active reclamation activities. Additional sampling of disturbed sites can also help refine ecosystem recovery indicator forecasts by disturbance type and increase the accuracy of the results derived. This research utilized only 10 control sites to derive mean undisturbed conditions, so additional sampling of undisturbed sites – ideally some on either side of each creek drainage sampled for post-mining attributes -- could also lead to further refinement of forecasts for ecosystem recovery. Data for disturbed sites that have been abandoned for longer periods of time were more difficult to locate, with the majority of the sites sampled (n=82) having less than 35 years since disturbance. A few aspects of the older sites existing in the region, including a lack of knowledge regarding their exact location, poor accessibility by road, and the reprocessing of older mine tailings with modern technologies somewhat limited sampling to younger sites. To better refine estimates for ecosystem recovery timelines, further sampling of sites >35 years since disturbance is recommended. Similarly, having more control plots around each creek drainage would help further refine control means, and further refine the estimates made for time needed for ecosystem recovery to a selected endpoint. Due to the size of sample plots, limited sampling ability, and limited experience, it is very difficult to infer any causation to the wildlife activity data collected. Wildlife 57 activity on each plot was sampled a single time, with sites not revisited to check for new evidence of wildlife activity. This creates the uncertainty in regard to whether the plot was being actively used by animals, or whether the site just being passed through and not being used as active habitat (Morrison et al., 2008). Continued monitoring is suggested for better insight on active habitat use. Furthermore, research into the effects of placer mining activities on nearby wildlife habitat use was not conducted for currently operating placer mining locations; conducting further research on aspects such as the effects of human presence and noise generated by operations on wildlife patterns is recommended (Blickey et al., 2012). Similarly, due to limited experience with plant species identification, the accuracy of plant cover by species and subsequent calculations of species richness and compositional similarity must be considered very coarse. Further refinement to species identification (especially with respect to bryophytes and lichens) and species richness calculations could help refine estimates made for the amount of time needed for ecosystem recovery and of its governing factors. 58 References cited Alavalapati, J. R., and Adamowicz, W. L. 2000. Tourism impact modeling for resource extraction regions. Annals of Tourism Research, 27(1), 188-202. Anderson, D. R., Laake, J. L., Crain, B. R., and Burnham, K. P. 1979. Guidelines for line transect sampling of biological populations. The Journal of Wildlife Management, 43(1), 70-78. Ball, D. F. 1964. Loss-on-Ignition as an estimate of organic matter and organic carbon in non-calcareous soils. Journal of Soil Science, 15, 84–92. Banerjee, A., Chitnis, U., Jadhav, S., Bhawalkar, J., and Chaudhury, S. 2009. Hypothesis testing, type I and type II errors. Industrial Psychiatry Journal, 18(2), 127–131. Available on-line at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2996198/ [Accessed Mar. 30 2017] Banner, A., MacKenzie, W., Haeussler, S., Thomson, S., Pojar, J., and Trowbridge, R. 1993. A field guide to site identification and interpretation for the Prince Rupert Forest Region. B.C. Min. For., Res. Br., Victoria, B.C., Land Manage. Handbook. No. 26. Available on-line at: http://www.for.gov.bc.ca/hfd/pubs/docs/lmh/lmh26.htm [Accessed 14 Nov. 2016] BCMFLNOEM (BC Ministry of Forests, Lands and Natural Resource Operations and Ministry of Energy and Mines). 2014. Atlin Placer Mining Best Management Practices Guidebook. Available online at: http://www.elp.gov.bc.ca/wld/documents/bmp/Skeena/Atlin Placer Mining BMP Guidebook_FINAL June 30 2014.pdf [Accessed 14 Nov. 2016]. BCMFRE (BC Ministry of Forests and Range and BC Ministry of Environment). 2010. Field Manual for Describing Terrestrial Ecosystems, 2nd Edition. B.C. Ministry of Forests and Range and B.C. Ministry of the Environment, Victoria, B.C. BCTS (Blue Canyon Technical Subgroup). n.d. Interim Blue Canyon Subgroup Report: Draft. Taku River Tlingit First Nation, B.C. Ministry of Energy, Mines and Natural Gas, and B.C. Ministry of Forests, Lands and Natural Resource Operations. 8 p. BCTRTFN (Province of British Columbia and Taku River Tlingit First Nation). 2011. Wóoshtin wudidaa: Atlin Taku Land Use Plan. July 19, 2011. Taku River Tlingit First Nation, Atlin, B.C., and Province of British Columbia, Victoria, B.C. 117 p. Available on-line at: https://www.for.gov.bc.ca/tasb/slrp/pdf/srmp/ATLIN-TAKU-LUP.pdf [Accessed 14 Nov. 2016]. Blickey, J. L., Blackwood, D., and Patricelli, G. L. 2012. Experimental evidence for the effects of chronic anthropogenic noise on abundance of greater sage-grouse at leks. Conservation Biology, 26(3), 461-471. Bradshaw, A.D., and Chadwick, M. J. 1980. The Restoration of Land: The Ecology and Reclamation of Derelict and Degraded Land. Blackwell, London. 317 p. 59 Brouillet L, Desmet P, Coursol F, Meades SJ, Favreau M, Anions M, Bélisle P, Gendreau C, Shorthouse D, and contributors 2010+. Database of Vascular Plants of Canada (VASCAN). released on 2010-12-10. Available on-line at: Online at http://data.canadensys.net/vascan [Accessed 14 Nov. 2016] Burton, P. J., and Haig, J. B. 2016. Assessing and Promoting Ecosystem Recovery After Placer Mining in the Atlin Area. Report to BC Ministry of Forests, Lands and Natural Resource Operations, Skeena Region. Natural Resources and Environmental Studies, University of Northern B.C., Prince George, B.C. 51 p. CDNRMS (Colorado Department of Natural Resources, Division of Reclamation Mining & Safety). 1988. Guidelines for Compliance with Land Use And Vegetation Requirements of the Colorado Mined Land Reclamation Board for Coal Mining. Available on-line at: http://mining.state.co.us/Programs/Coal/RulesRegs/Documents/1988VegetationG uidelines.pdf [Accessed 20 Nov. 2016] Cederlund, G., and Sand, H. 1994. Home-range size in relation to age and sex in moose. Journal of Mammalogy, 75(4), 1005-1012. Christensen, A. F., He, H., Dyck, M. F., Turner, L. E., Chanasyk, D. S., Naeth, M. A., and Nichol, C. 2013. In situ measurement of snowmelt infiltration under various topsoil cap thicknesses on a reclaimed site. Canadian Journal of Soil Science, 93(4), 497510. Cooper, D. J., and Van Haveren, B. P. 1994. Establishing felt-leaf willow from seed to restore Alaskan, USA, floodplains. Arctic and Alpine Research 26(1), 42-45. De'ath, G., and Fabricius, K. E. 2000. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology 81(11), 3178-3192. de Groot, A., and Pojar, J. 2009. Sensitive Ecosystems of the Atlin-Taku Planning Area. Bulkley Valley Centre for Natural Resources Research and Management, Smithers, B.C. 66 p. Densmore, R. 1994. Succession on regraded placer mine spoil in Alaska, U.S.A., in relation to initial site characteristics. Arctic and Alpine Research, 26(4), 354-363. Densmore, R.V. 2005. Succession on subalpine placer mine spoil: Effects of revegetation with Alnus viridis, Alaska, USA. Arctic, Antarctic, and Alpine Research, 37(3), 297303. Dickinson, C.F., and Smith, D. S. 1995. Atlin: The Story of British Columbia’s Last Gold Rush. Atlin Historical Society, Atlin, B.C. 379 p. Dowling, P. M., Clements, R. J., and McWilliam, J. R. 1971. Establishment and survival of pasture species from seeds sown on the soil surface. Crop and Pasture Science, 22(1), 61-74. Entrix, I. 1986. Best Management Practices for Placer Mining. 1st ed. Alaska Department of Fish and Game. Available at: 60 https://www.adfg.alaska.gov/static/home/library/pdfs/habitat/placer_mining_bes t_practices.pdf[Accessed 14 Nov. 2016]. Evans, R. and Young, J. 1987. Seedbed Microenvironment, Seedling Recruitment, and Plant Establishment on Rangelands. 1st ed. [ebook] Available on-line at: http://db.lib.uidaho.edu/ereserve/courses/r/range/440_01/art08.pdf [Accessed 14 Nov. 2016]. Farrington, J. 2000. Environmental problems of placer gold mining in the Zaamar Goldfield, Mongolia. World Placer Journal, 1, 107-128. Gee, G. W., & Or, D. 2002. Particle-size analysis. Methods of soil analysis. 4, 255-293. Gretarsdottir, J., Aradottir, A. L., Vandvik, V., Heegaard, E., and Birks, H. J. B. 2004. Long‐ term effects of reclamation treatments on plant succession in Iceland. Restoration Ecology, 12(2), 268-278. Green, J.E., Van Egmond, T.D., Wylie, C., Jones, I., Knapik, L. and Paterson, L. R. 1992. A User Guide to Pit and Quarry Reclamation in Alberta. Reclamation Research Technical Advisory Committee, Alberta Land Conservation and Reclamation Council. Edmonton, Alberta. 137 p. Griffin, B. 1992. Miners at work, a history of British Columbia’s gold rushes. Pioneering Geology in the Canadian Cordillera: British Columbia Geological Survey, Open File, 19, 5-18. Hagler Bailly Inc. 1998. Best Management Practices in Nonferrous Metals Mining and Processing [Scholarly project]. In United States Agency of International Development. Available on-line at: http://pdf.usaid.gov/pdf_docs/pnaeb046.pdf [Accessed 14 Nov. 2016] Halvorson, G. A., Melsted, S. W., Schroeder, S. A., Smith, C. M., and Pole, M. W. 1986. Topsoil and subsoil thickness requirements for reclamation of nonsodic minedland. Soil Science Society of America Journal, 50(2), 419-422. Hamrick, J. L., and Lee, J. M. 1987. Effect of soil surface topography and litter cover on the germination, survival, and growth of musk thistle (Carduus nutans). American Journal of Botany, 451-457. Hardy Associates. 1978. Fish and wildlife habitat recovery in placer mined areas of the Yukon. Final Report. Prepared for Dept. of Indian Affairs and Northern Development. Calgary, Alberta. Harestad, A. S., and Bunnell, F. L. 1979. Home range and body weight‐‐A reevaluation. Ecology, 60(2), 389-402. Harper, J. L., Williams, J. T., and Sagar, G. R. 1965. The behaviour of seeds in soil: I. The heterogeneity of soil surfaces and its role in determining the establishment of plants from seed. Journal of Ecology, 53(2), 273-286. 61 Hayter, R. 2000. Single Industry Resource Towns. In E, Sheppard and T. Barnes (eds) A Companion to Economic Geography Oxford: Basil Blackwell, 290-307. Kennedy, R. J. 1969. Sunbathing behaviour of birds. British Birds, 62(7), 249-258. Laarmann, D., Korjus, H., Sims, A., Kangur, A., Kiviste, A., and Stanturf, J. A. 2015. Evaluation of afforestation development and natural colonization on a reclaimed mine site. Restoration Ecology, 23(3), 301-309. Landres, P. B., Morgan, P., & Swanson, F. J. 1999. Overview of the use of natural variability concepts in managing ecological systems. Ecological Applications, 9(4), 1179-1188. Idaho Department of Lands. n.d. Dredge and Placer mining operations in Idaho. Idaho Department of Lands, Boise, Idaho. Available on-line at http://adminrules.idaho.gov/rules/current/20/0301.pdf [Accessed 14 Nov. 2016] Johnson, L. 1987. Management of northern gravel sites for successful reclamation: A review. Arctic and Alpine Research, 19(4), 530-536. doi:10.2307/1551421 LaPerriere, J. D. and Reynolds, J. B., 1997. Gold placer mining and stream ecosystems of interior Alaska. Pages 265-280 in A. M. Milner and M. W. Oswood, editors. Freshwaters of Alaska: Ecological Syntheses. Springer-Verlag, New York. Magurran, A.E. 1988. Ecological Diversity and its Measurement. Princeton University Press, Princeton, New Jersey. 179 p. Måren, I. E., Karki, S., Prajapati, C., Yadav, R. K., and Shrestha, B. B. 2015. Facing north or south: Does slope aspect impact forest stand characteristics and soil properties in a semiarid trans-Himalayan valley. Journal of Arid Environments, 121, 112-123. McCarthy, D.P., Luckmann, B. H., and Kelly, P. E. 1991. Sampling height–age error correction for spruce seedlings in glacial forefields, Canadian Cordillera. Arctic and Alpine Research, 23, 451–455. McCune, B., Grace, J. B., and Urban, D. L. 2002. Analysis of Ecological Communities. MjM Software, Gleneden Beach, Oregon, U.S.A. 300 p. McCune, B., and Mefford, M. J. 2011. PC-ORD: Multivariate Analysis of Ecological Data, Version 6. MjM Software, Gleneden Beach, Oregon, U.S.A. Miller, R. M., and Jastrow, J. D. 1992. The application of VA mycorrhizae to ecosystem restoration and reclamation. Pages 488-517 in M.F. Allen, Chapman, and Hall, editors. Mycorrhizal Functioning: An Integrative Plant-Fungal Process, Springer Science & Business Media, New York. Miller, R. M., Carnes, B. A., & Moorman, T. B. 1985. Factors influencing survival of vesicular-arbuscular mycorrhiza propagules during topsoil storage. Journal of Applied Ecology, 22(1), 259-266. 62 Morrison, M. L., Block, W. M., Strickland, M. D., Collier, B. A., and Peterson, M. J. 2008. Wildlife study design. Springer Science & Business Media, New York. NAIT (Northern Alberta Institute of Technology). 2015. Progressive Reclamation. 1st ed. Technical Note. [ebook] Northern Alberta Institute of Technology, Edmonton, AB. Available on-line at: http://www.nait.ca/docs/1_progressive_reclamation.pdf [Accessed 14 Nov. 2016]. Oades, J. M. 1984. Soil organic matter and structural stability: mechanisms and implications for management. Plant and soil, 76(1-3), 319-337. Pickett, S. T. 1989. Space-for-time substitution as an alternative to long-term studies. Pages 110-135 in Likens, G. Springer, editor. Long-Term Studies in Ecology. New York. Pimentel, D., & Kounang, N. (1998). Ecology of soil erosion in ecosystems. Ecosystems, 1(5), 416-426. Polster, D. 2011. Towards Revegetation Sustainability Criteria for Northern Mine Closure. Report prepared for Independent Environmental Monitoring Agency, Yellowknife, Northwest Territories. 18 p. Available on-line at http://www.monitoringagency.net/LinkClick.aspx?fileticket=AcA9Q8u0fcM%3D&t abid=110 [Accessed 14 Nov. 2016]. Prach, K., and Hobbs, R. J. 2008. Spontaneous succession versus technical reclamation in the restoration of disturbed sites. Restoration Ecology, 16(3), 363-366. R Core Team. 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.Rproject.org/. [Accessed 14 Nov. 2016]. Rieger, J., Stanley, J. and Traynor, R. 2014. Project Planning and Management for Ecological Restoration. Island Press, Washington, DC. 300 p. Rhoades, C., Binkley, D., Oskarsson, H. and Stottlemyer, R. 2008. Soil nitrogen accretion along a floodplain terrace chronosequence in northwest Alaska: influence of the nitrogen-fixing shrub Shepherdia canadensis. Ecoscience 15(2), 223-230. Römkens, M. J., Helming, K., and Prasad, S. N. 2002. Soil erosion under different rainfall intensities, surface roughness, and soil water regimes. Catena, 46(2), 103-123. Ruiz-Jaén, M.C., and T.M. Aide T.M. 2005. Vegetation structure, species diversity, and ecosystem processes as measures of restoration success. Forest Ecology and Management 218(1), 159-173. Russell, E.W. 1973. Soil Conditions and Plant Growth. Longman Group Limited, New York. 849 p. Schulte, E. E., Kaufmann, C., and Peter, J. B. 1991. The influence of sample size and heating time on soil weight loss on ignition. Communications in Soil Science & Plant Analysis, 22(1-2), 159-168. 63 SER (Society for Ecological Restoration). 2004. The SER International Primer on Ecological Restoration, Version 2. Science and Policy Working Group, Society for Ecological Restoration International, Tucson, Arizona. Available on-line at: http://c.ymcdn.com/sites/www.ser.org/resource/resmgr/custompages/publicatio ns/SER_Primer/ser_primer.pdf?hhSearchTerms=%22primer%22 [Accessed 18 Nov. 2016]. Sheoran, V., Sheoran, A.S. and Poonia, P. 2010. Soil reclamation of abandoned mine land by revegetation: A review. International Journal of Soil, Sediment and Water 3(2): Article 13. Available on-line at: http://scholarworks.umass.edu/intljssw/vol3/iss2/13 [Accessed 14 Nov. 2016]. Singh, R. S., Tripathi, N., and Hills, C. D. 2016. Reclamation of Mine-impacted Land for Ecosystem Recovery. John Wiley & Sons. SMEGAC (The Saskatchewan Mineral Exploration and Government Advisory Committee). 2012. Mineral Exploration Guidelines for Saskatchewan. Saskatchewan Department of the Environment, Regina, SK. Available online at: http://www.environment.gov.sk.ca/mineralexploration [Accessed 14 Nov. 2016]. Statistics Canada. 2012. Stikine Region, British Columbia (Code 5957022) and British Columbia (Code 59) (table). Census Profile, 2011 Census. Statistics Canada Catalogue no. 98-316, October 24, 2012. Steinberg, D. n.d.. R-Squared for CART Regression Trees. Available on-line at http://blogs.salford-systems.com/dan-steinberg/r-squared-for-cart-regressiontrees [accessed 14 Nov, 2016] Therneau, T., Atkinson, B. and Ripley, B. 2015. rpart: Recursive Partitioning and Regression Trees. R package version 4.1-10. Available on-line at: https://cran.rproject.org/web/packages/rpart/index.html [Accessed 14 Nov. 2016] Thurber Consultants Ltd., Land Resources Network Ltd., and Norwest Soil Research Ltd. 1990. Review of the Effects of Storage on Topsoil Quality. Report No. RRTAC 905. Alberta Land Conservation and Reclamation Council. Edmonton, Alberta. 116 p. Trowbridge, R. 2014. Reconnaissance Sampling of Ecosystem Recovery after Placer Mining in the Atlin Area. Contract Summary Report to BC Ministry of Forests, Lands and Natural Resource Operations, Smithers, B.C. Contract. File No: 1000540/CS1582002. TRTFN (Taku River Tlingit First Nation). 2013. Proposed Model for Assessment of Disturbance Class for Placer Tenures in Blue Canyon / At Xa Koogu ASRMZ. Draft, July 15, 2013. Taku River Tlingit First Nation, Atlin, B.C. 7 p. TRTFNBC (Taku River Tlingit First Nation and Province of British Columbia). Wóoshtin Yan Too.Aat: Land and Resource Management and Shared Decision Making 64 Agreement Between the Taku River Tlingit First Nation and the Province of British Columbia. Taku River Tlingit First Nation, Atlin, B.C., and Prince of British Columbia, Victoria, B.C. 44 p. Available on-line at: http://www.newrelationship.gov.bc.ca/shared/downloads/signed_TRTFN_BC_LR MSDM_agreement_.pdf [Accessed 14 Nov. 2016]. Walker, L. R., Walker, J., and Hobbs, R. J., Eds. 2007. Linking Restoration and ecological succession Springer, New York, NY. Walker, L.R., Zasada, J.C. and Chapin, F.S. III. 1986. The role of life history processes in primary succession on an Alaskan floodplain. Ecology, 67, 1243–1253 Western Ag Innovations Inc. 2013. Plant Root Simulator (PRS) Probes. Western Ag Innovations Incorporated, Saskatoon, Saskatchewan. Available on-line at: https://www.westernag.ca/innovations/technology/basics [Accessed 14 Nov. 2016] Yukon Tourism and Culture 2010. Yukon Placer Mining Best Management Practices for Heritage Resources. Whitehorse, Yukon. Available on-line at: http://www.tc.gov.yk.ca/pdf/Placer_Mining_BMP.pdf [Accessed 14 Nov. 2016]. Yukon Regulations. 2013. Placer Mining Land Use Regulation. The Government of Yukon Department of Energy, Mines and Resources. Whitehorse, Yukon. Avaliable on-line from http://www.gov.yk.ca/legislation/regs/oic2003_059.pdf [Accessed 14 Nov. 2016] 65 Appendix I. Field data collection form. 66 (Appendix I. Field data collection form, continued) 67 Appendix II. Site and vegetation data used in statistical analysis. 68 Appendix II. Site and vegetation data, continued. 69 Appendix II. Site and vegetation data, continued. 70 Appendix II. Site and vegetation data, continued. 71 Appendix III. Plant species frequency (%) on areas disturbed by placer mining (n=90) and nearby undisturbed (n=10) “control” sites east of Atlin Lake, B.C. Tree Species Abies lasiocarpa Picea glauca Picea mariana Pinus contorta Populus balsamifera Populus tremuloides Salix scouleriana Disturbed Sites 15.6 37.8 0.0 24.4 57.8 3.3 13.3 Control Sites 20 50 10 40 0 20 20 Shrub Species Alnus viridis Arctostaphylos uva-ursi Betula glandulifera Dryas drummondii Empetrum nigrum Juniperis communis Ledum groenlandicum Linnaea borealis Potentilla fruticosa Ribes laxiflorum Rosa acicularis Rubus arcticus Rubus idaeous Salix alaxensis Salix candida Salix glauca Salix sp. Sheperdia canadensis Vaccinium vitis-idaea Disturbed Sites 1.1 12.2 5.6 6.7 2.2 6.7 1.1 0.0 4.4 2.2 0.0 7.8 2.2 61.1 51.1 10.0 3.3 42.2 2.2 Control Sites 0 20 30 0 30 20 30 40 0 20 30 10 0 30 0 0 0 40 50 Grass and Grass-like Spp. Agropyron alaxensis Agrostis sp. Agrostis scabra Calamagrostis sp. Carex sp. Carex media Deschampsia caespitosa Disturbed Sites 33.3 4.4 2.2 5.6 2.2 10.0 7.8 Control Sites 20 10 0 10 10 10 10 72 Grass and Grass-like Spp. Eriophorum chamissonis Festuca saximontana Hordeum jubatum Juncus arcticus Juncus drummondii Luzula parviflora Phleum alpinum Poa sp. Poa alpina Typha latifolia Trisetum spicatum Disturbed Sites 2.2 0.0 5.6 2.2 5.6 0.0 2.2 13.3 5.6 1.1 10.0 Control Sites 0 0 10 0 10 0 0 10 10 0 10 Broadleaf Herbs Achillea millifolium Anemone multifida Antennaria microphylla Arnica cordifolia Artemisia norvegica Astragalus alpinus Blitum capitatum Castilleja unalaschcensis Comandra sp. Cornus canadensis Delphinium glaucum Epilobium angustifolium Epilobium ciliatum Epilobium latifolium Erigeron sp. Eurybia sibirica Galium boreale Geum triflorum Hedysarum boreale Heracleum maximum Hieracium gracile Lupinus arcticus Mertensiana paniculata Myosotis alpestris Nasturtium officianale Orthillia secunda Oxytropis campestris Broadleaf Herbs Disturbed Sites 53.3 5.6 10.0 5.6 6.7 8.9 1.1 1.1 1.1 6.7 5.6 47.8 0.0 34.4 6.7 3.3 3.3 2.2 6.7 1.1 15.6 24.4 16.7 1.1 1.1 3.3 13.3 Disturbed Sites Control Sites 60 0 20 10 0 10 0 0 0 0 0 70 0 20 10 0 0 0 0 0 10 20 20 0 0 0 0 Control Sites 73 Penstemon procerus Polemonium pulcherrimum Potentilla gracilis Ranunculus sp. Sanguisorba sitchensis Senecio pauperculus Solidago multiradiata Taraxacum officinale Trifolium hybridum Tussilago farfara Veronica sp. Fern-Allies Equisetum arvense Equisetum fluviatale Equisetum sylvaticum Lycopodium annotinum Lycopodium complanatum Bryophytes Aulocomnium palustre Brachythecium sp. Ceratodon purpureus Dicranum sp. Hylocomium splendens Marchantia polymorpha Pleurozium schreberi Polytrichum juniperinum Ptilium crista-castrensis Lichens Cladina rangiferina Cladonia cariosa Cladonia cornuta Cladonia pyxidata Cladonia sp. Flavocetraria cucullata Peligera sp. Peltigera apthosa Stereocaulon paschale 2.2 6.7 3.3 0.0 12.2 0.0 12.2 45.6 2.2 2.2 1.1 Disturbed Sites 34.4 6.7 0.0 0.0 1.1 Disturbed Sites 4.4 17.8 14.4 0.0 5.6 0.0 4.4 3.3 4.4 Disturbed Sites 3.3 0.0 2.2 5.6 1.1 0.0 1.1 8.9 3.3 0 10 0 0 10 0 10 70 0 0 0 Control Sites 40 0 0 0 0 Control Sites 0 0 0 0 10 0 10 0 10 Control Sites 0 0 0 0 0 0 0 10 10 74 Appendix IV. Additional classification and regression trees. Figure IV-1. Classification and regression tree for factors influencing compositional similarity to control sites on post-mining sites east of Atlin Lake, B.C. R 2 = 0.46. 75 Figure IV-2. Classification and regression tree for factors influencing structural diversity (H’) on post-mining sites east of Atlin Lake, B.C. R2 = 0.51. 76 Figure IV-3. Classification and regression tree for factors influencing A horizon depth (cm) on post-mining sites east of Atlin Lake, B.C. R2 = 0.30. 77 Figure IV-4. Classification and regression tree for factors influencing wildlife species richness on post-mining sites east of Atlin Lake, B.C. R2 = 0.63. 78 Appendix V. Significant differences in ecosystem recovery indicators among categorical subsets of the data, based on ANOVA results. Factor Indicator Creek Vascular spp. drainage Richness BEC Zone Structural diversity, H’ BEC Zone Vascular spp. Richness, count BEC Zone Wildlife activity, sign count Slope position A horizon depth, cm Disturbance Structural diversity, type H’ Disturbance A horizon depth, cm type Reclamation A horizon depth, cm done Soil texture Summed vegetation group cover, % F-value F13,76 =2.02 F1,88 =6.50 pSignificant differences value 0.0297 Tukey n.s. 0.0125 BWBS (1.055) > SWB (0.883) F1,88 =5.10 0.0264 BWBS (9.9) > SWB (8.2) F1,88 =4.01 0.0484 BWBS (2.9) < SWB (4.7) F5.84 =2.33 F5,84 =2.89 F2,87 =6.11 0.0491 Tukey n.s. 0.0186 OBWP (1.13) ≥ RD (1.02) = STA (0.98) = SP (0.95) ≥ WSGC (0.62) = RCPW (0.61) 0.0297 OBWP (8.0) ≥ SP (3.9) = STA (1.8) = RD (1.5) ≥ RCPW (0) = WSGC (0) 0.0033 PB (8.) ≥ CON (4.5) ≥ NAT (1.3) F3,86 =5.01 0.0030 Loam (78) > silt (49) = sand (47) > clay (36) F5,84 =2.62 BWBS = boreal white and black spruce biogeoclimatic zone, SWB = spruce-willow-birch biogeoclimatic zone OBWP = overburden waste pile; RD = rock dump; STA = scraped traffic area; SP = settling pond; WDGC = washed sand/gravel/cobbles RCPW = road cut, pitwalls PB = soil and organics pulled back; CON = contoured or re-sloped; NAT = left for natural recovery 79 Appendix VI. Significant (p<0.05) correlations of site, soil, and vegetation factors with indicators of ecosystem recovery. Associated Summed Variables Vegetation Cover Easting Northing Elevation Slope Microsite relief Soil 0.42 moisture regime Soil 0.51 nutrient regime Time since disturbance Coarse fragments Tree cover, >10m tall Tree cover, 0.39 >5m tall Shrub layer 0.41 cover Herb layer cover Moss layer 0.51 cover Dead wood 0.22 cover Rock cover -0.28 Bare soil -0.69 cover Soil 0.26 looseness Coarse soil -0.24 Structural Vascular Compositional Diversity Species Similarity to Richness Control -0.22 -0.22 -0.24 Wildlife “A” Activity horizon depth 0.45 0.29 -0.22 -0.31 -0.29 -0.26 0.28 0.31 0.54 0.21 0.22 -0.27 0.46 0.46 0.44 0.36 0.24 0.23 0.25 0.21 0.61 0.33 0.25 -0.24 -0.21 -0.35 80 fraction Medium soil fraction Fine soil fraction pH Surface erosion cover Nitrate N flux Ammonium N flux Calcium cation flux Magnesium cation flux Potassium cation flux Phosphate anion flux Iron cation flux Copper cation flux Zinc cation flux 0.24 0.30 -0.22 -0.22 -0.21 -0.27 -0.23 -0.26 0.29 0.25 -0.26 -0.22 0.57 0.28 -0.28 -0.33 0.39