Passive Soil Manipulation Influences the Successional Trajectories of Forest Communities at a Denuded Former Campsite

©2012 by the Board of Regents of the University of Wisconsin System. Passive Soil Manipulation Influences the Successional Trajectories of Forest Communities at a Denuded Former Campsite Nate Hough-Snee, University of Washington School of Forest Resources and Center for Urban Horticulture, Box 354115, Seattle, WA 98195-2100, USA; Present address: Utah State University Ecology Center and Department of Watershed Sciences, 5210 Old Main Hill, Logan, UT 84322-5210, USA; A. Lexine Long, University of Washington School of Forest Resources and Center for Urban Horticulture, Box 354115, Seattle, WA 98195-2100, USA; Present address: Utah State University Ecology Center and Department of Watershed Sciences, 5210 Old Main Hill, Logan, UT 84322-5210, USA; Rodney Pond (corresponding author), University of Washington School of Forest Resources and Center for Urban Horticulture, Box 354115, Seattle, WA 98195-2100, USA, rodney.pond@gmail.com.

P assive site manipulation is an ecological restoration approach that strives to improve heavily degraded sites by altering physical conditions to encourage the establishment of autogenic ecosystem processes. Passive restoration approaches have been successfully used in numerous ecosystems to alter soil nutrition, moisture retention, and propagule retention, which can drive both the establishment of vegetation and plant successional trajectories (Ewing et al. 2002, Cole 2007, Hough-Snee et al. 2011. In some ecosystems, increasing plant recruitment from seed rain and/or intentional seeding through microsite enhancement may provide an effective low-investment, low-intensity revegetation strategy to land managers when unmediated, 'natural' revegetation is unfeasible due to poor site conditions (Cole 2007). One application of passive restoration techniques is in heavily impacted sections of natural areas where the surrounding vegetation matrix is strongly intact but ecosystem processes may act too slowly to reclaim the disturbed site or prevent further degradation. Active planting approaches may be logistically difficult or prohibitively expensive in some natural areas, leaving passive site manipulation as the only feasible option. In 2006, we used low-investment surface soil treatments to retire a denuded former campsite and restore the resulting disturbed area to local forest vegetation. This project was an opportunity to test the effectiveness of 3 seedbed treatments in recruiting tree species from seed rain and to concurrently assess forest ecosystem development over time.
Our project site is adjacent to the North Cascades Environmental Learning Center (ELC) on the northern shore of Diablo Lake in the North Cascades National Park Complex (Marblemount, Washington, USA). The ELC was developed on the site of a former fishing camp where after camp removal in 2004, the site was used for construction material storage, resulting in highly compacted soils, denuded entirely of vegetation and soil organic matter. The surrounding forest occurs in the northern reach of Washington's western hemlock (tsuga heterophylla) zone (Franklin and Dyrness 1988) and is dominated by western hemlock, Douglas-fir (Pseudotsuga menziesii), lodgepole pine (Pinus contorta var. latifolia), red alder (alnus rubra), bigleaf maple (acer macrophyllum), and paper birch (Betula papyrifera). Forest canopy near the ELC ranges from 70-90% cover in most areas, with the largest canopy gaps occurring adjacent to roads. The ELC is situated at an elevation of ≈ 370 m with a southwest aspect and slopes ranging from 3-5%. Outside of the immediate 1-ha restored area, surrounding forest hydrology, soils and vegetation have remained largely undisturbed.
We anticipated that the compacted ELC site would benefit from soil treatments of different source materials and particle size classes that promoted heterogeneous seedling establishment across the site. More specifically, we hypothesized that the dominant tree species establishing within each treatment would correspond to the soil treatments most conducive to their seed size and type. We used a completely randomized block design to evenly allocate 3 soil treatments, scarified mineral soil, forest litter, and woodchip mulch, to 120 1m×1m plots across 3 blocks. The scarified soil treatment (S) consisted of post-denudation mineral soils ( particle size <0.5cm) free of organic matter that were raked and spread evenly across each plot. The forest litter treatment (FL) contained needles, seed cones, lichens, and small twigs collected from accumulations on nearby maintained campsites ( particle size ≈1-3cm). We spread litter evenly over scarified denuded mineral soil to a depth of ≈2 cm. The woodchip mulch treatment (WM) was comprised of shredded red alder wood chips ( particle size ≈2-5 cm) that were applied directly to existing scarified mineral soil at a depth of 2 cm (Table 1). We monitored natural vegetation recruitment into these sites in 2007 and 2010 by collecting tree and shrub abundance data, counting individual stems of all woody species within each plot, measuring woody species stem height, and estimating percent cover of herbaceous vegetation and bare ground. In 2007, total sample size was reduced from 120 plots to 101 plots when a fallen tree destroyed 19 plots. Because our objective was to identify patterns in potential canopy tree species, and herbaceous cover commonly totaled <5%, we excluded herbaceous vegetation from our analyses. We used PERMANOVA, a permutation-based ANOVA test that uses distance matrices to assess differences between community groups (Anderson 2001), to compare the vegetation communities between treatments and to perform between-group pairwise comparisons. To illustrate what species drove differences in community composition, we used indicator species analysis (ISA; Bakker 2008) to define species indicative of each treatment. ISA calculates the product of each species' relative frequency and relative abundance in a given treatment and returns an indicator value between 0 and 100. As a species' indicator value approaches 100, it is thought to be a more perfect indicator of a given treatment. Each species' indicator value is statistically assessed using Monte Carlo randomizations to return a p-value. We used 10,000 permutations, constraining permutations within each block, to derive p-values for both PERMANOVA and ISA analyses. We performed analyses independently for years 2007 and 2010 using R statistical software (version 2.11.1) and considered a species' indicator value (IV) to be statistically significant and PERMANOVA groups to significantly differ if p < 0.05. Seedbed manipulation to promote seed recruitment and seedling establishment has historically served as a primary tool to overcome thresholds for reinitiating ecosystem processes on damaged sites (Luken 1990, Whisenant 1999. At ELC we observed that when intentional seeding proves to be impractical and the damaged site occurs within an intact matrix of reproductively mature species, applying specific seedbed treatments to select for and promote desired species offers a viable passive restoration approach that can reinitiate ecosystem processes and set early-successional forest trajectories. All 3 treatments clearly demonstrated that enhancing soil surface roughness successfully recruits canopy and woody understory species from seed rain (Table  1). Simple soil scarification, either as a single treatment or when combined with light applications of either woodchip mulch or forest litter, retained seed, promoted germination, and allowed for plant establishment and survival 3 yr after application. Furthermore, varying soil surface treatments proved to be strongly selective and significantly drove species composition both 1 and 4 yr following restoration.
The change in indicator species over a brief 3-yr period suggests that seedbed conditions exert a strong selective pressure during colonization but interact with environmental filters (light availability, herbivory, soil moisture and temperature, etc.) and species performance to determine which species persist over time. We expect that highly shade intolerant red alder (mean 2010 height = 81.92 cm), though the most common recruit in the S treatment and overall, will not be recruited into the overstory canopy because the experimental blocks occupy small (< 500 m2) canopy gaps. Similarly, shade intolerant lodgepole pine (mean 2010 height = 19.06 cm) and Douglas-fir (mean 2010 height = 11.04 cm) may be suppressed by the taller, more densely colonizing red alder ( Figure 1). Given the existing canopy gaps in the experimental blocks, we expect that shade tolerant western redcedar and western hemlock are the most likely species to persist long enough to reach the canopy. Subsequently, other factors such as ungulate herbivory, local climate and fire cycles, and pathogen outbreaks may further determine which recruited species reach the forest canopy. We conclude that selecting a lowinvestment seedbed preparation to recruit target species from seed rain must be done with both the knowledge of target species seedbed preferences and acknowledgement that initial treatments may not allow recruited species to persist to maturity. Additional treatments may need to be undertaken to shape selective forces that enhance the likelihood that target species will persist to maturity and continue to drive forest vegetation communities. F or decades, Madagascar has drawn scientific attention for its unique biodiversity and high levels of endemism. This island nation has long been recognized for its remarkable biota, which has evolved during nearly 100 million yr of tectonic isolation, with a diverse range of climates and substrates, and a varied topography (Kull 2000). Current estimates suggest that the Malagasy flora consists of approximately 13-14,000 species, more than 90% endemic (Phillipson et al. 2006), which occur in a wide range of native vegetation types, many of which have been highly degraded or replaced by secondary formations resulting from human activities (Lowry et al. 1997).

Ibity Mountain, Madagascar: Background and Perspectives for Ecological Restoration
Tapia woodland is a sclerophyllous formation restricted to Madagascar's central highlands, dominated by the endemic tapia tree (Uapaca bojeri). This formation is adapted and resilient to the region's natural fire regime and occurs today in scattered, isolated stands that total no more than 132,255 ha. Most other native woody vegetation types of the central highlands have been replaced by grasslands formed during the last 1,500-2,000 yr primarily by a human-caused increase in fire frequency well above that of the natural regime. These grasslands are burned annually to renew livestock pasture, clean fields, or control pests (Kull 2000), and fires encroach well into the interior of the tapia woodland. Today the remaining tapia woodland has an extensive herbaceous layer and an open canopy with many woody representatives of the endemic family Sarcolaenaceae (Lowry et al. 1997, Figure 1). Traditional management practices appear to have spared remnants of tapia woodland because this vegetation plays an important role in the local economy as a source of fuel wood, edible fruits, mushrooms and medicinal plants, and as habitat for wild silk worms (Borocera madagascariensis), whose cocoons are collected by local residents to produce cloth (Gade 1985, Kull 2002. However, the remaining stands are fragmented and structurally-modified by a changing fire regime that hinders woody species regeneration. Tapia woodland was once managed using traditional methods, which allowed sustainable extraction of forest resources for local consumption. Today it is still partly managed using traditional methods, but newer practices have caused degradation through increased cultivation and grazing, illegal exploitation and marketing of natural resources, reforestation with exotic tree species such as swamp mahogany (Eucalyptus robusta), Baguio pine (Pinus • 13 kesiya) and silver wattle (acacia dealbata), and changes in the fire regime.
A new protected area (PA) is currently being established at Ibity Massif (classified as type V, Harmonious Landscape, within the IUCN PA classification system). In December 2008, a temporary protection order was issued for Ibity (renewed in December 2010), and definitive establishment of the new PA is pending. A community-based conservation and restoration project is being coordinated by the Missouri Botanical Garden (MBG-Madagascar Program, Antananarivo) in cooperation with key Malagasy institutions and local communities. Ibity Massif is an approximately 6,000-ha quartzitic mountain located on the Malagasy highlands 25 km south of Antsirabe and 200 km south of the capital, Antananarivo. This massif is oriented north-south and ranges in elevation from 1400 to 2240 m. The MBG initiative has involved significant outreach and community education programs focusing on raising awareness of Ibity's conservation and economic importance, the threats to its biodiversity, and ongoing efforts to reduce fire frequency and implement ecological restoration projects with significant local community participation. This approximately 45-km2 PA contributes to the Madagascar Protected Areas System, which is scheduled to encompass over 6 million ha (about 10% of the country) (Borrini-Feyerabend and Dudley 2005). The Ibity PA will protect most of the massif 's characteristic environments (dense forest, tapia woodland and woody savanna, the latter two restricted to quartzitic substrates) and rare species.
On Ibity, lightening is thought to have been the main historical cause of wildfires, which are a natural part of the dynamics of many Malagasy ecosystems. However, the current human-caused high-frequency fire regime threatens the massif 's biodiversity and has reduced the extent of tapia woodland, fragmented the canopy, lowered the number of shrubs (Rabemanambola and Rakotoarisoa 2009), and increased the density of the herbaceous layer. Fires now commonly occur annually or biannually and burn both the herbaceous layer and part of the tree layer, although mature individuals in tapia woodland are somewhat resistant (Kull 2002). Young seedlings are, however, less fire tolerant and often die from burning. As a consequence, the shifting age structure in tapia woodland may ultimately lead to the disappearance of the tapia tree and its associated species.
In 2008, we began a research project to explore the relationship between fire and vegetation at Ibity, aiming to present an ecological characterization of this formation and an evaluation of the effects of fire on plant biodiversity. Identifying the conditions necessary for woody species regeneration will help to develop ecological restoration protocols for sclerophyllous woodlands. We are conducting: 1) an analysis of community composition and structure of Ibity's savanna and tapia woodlands; 2) a 2-yr study, initiated in 2010, on the phenology of woody and herbaceous taxa to assess species coexistence; and 3) an evaluation of seedling survival and establishment, and of woody species regeneration and survival, with the aim of providing a baseline restoration protocol for the reinforcement of woody plant populations in tapia woodland (Figure 2). This research program is designed to contribute to the management plan now being developed for the Ibity Massif, which will guide conservation activities once the site is formally designated as a new PA. The management plan is also being informed by discussions with stakeholders, including local community members, municipalities, regions, non-governmental organizations (NGOs), and the private sector. The plan aims to promote environmental protection, sustainable development, and the preservation of ethnobotanical relationships.
Sustainable development activities and careful environmental planning will be essential for the long-term viability of the new PA and for improving natural resource management. For Ibity's local communities, this project represents a new form of economic development because it provides much-needed employment opportunities, most particularly those related to eco-tourism. Several impending changes will lead to a reconfiguration of the area, requiring a balanced and equitable use of the landscape (Rabemanambola and Rakotoarisoa 2009). These include: a) the involvement in local sustainable development activities of a nearby cement factory operated since 1956 by HOLCIM, S.A. (a large Swiss-based company whose local headquarters is located in the town Antsirabe); and b) the pending establishment of the new Ibity PA, promoted since 2006 by local MBG personnel and stakeholders in the Vakinankaratra Region.
The creation of the Ibity PA and the implementation of an appropriate management plan will lead to local sustainable development through a consensus between those who seek protection and those who promote a standard development approach. This will require an appreciation among all stakeholders of the intimate relationships between culture and nature and between humans and the Earth, and a recognition of the aesthetic and scientific value of culturally important landscapes.  T he diversity in morphology of fishes is a product of the competitive processes that necessitate niche partitioning (MacArthur and MacArthur 1961). For example, interspecific competition is mediated through differentiation in mouth orientation; fishes with ventral mouths feed in the lower water column, whereas fishes with dorsal mouths feed in the upper water column (Wikramanayake 1990). Variation in water velocity reduces interspecific competition through character displacement, where low velocity waters contain fish with deep bodies and shorter fins and shallow, long-finned fishes are characteristic of high velocity waters (Wikramanayake 1990). Habitat heterogeneity facilitates the maintenance of morphological variation though an increase in dimensionality and habitat availability (MacArthur and MacArthur 1961). However, industrialization and development can strip lotic systems of heterogeneity. The Lower Humber River (LHR) watershed is entirely developed (TRCA 2008), with the LHR flowing from the Peel Plain through the City of Toronto and terminating at Lake Ontario. The LHR is surrounded by massive urban infrastructure; highway overpasses, rail lines, water treatment facilities, and residential high-rises surround the area. The watercourse is substantially altered to protect this infrastructure. There is extensive channelization resulting in the elimination of levee wetland systems and emergent and submergent aquatic vegetation that characterize natural estuaries. The channelization of the river has produced a single habitat type within the LHR, where fine grain size material and soft sediments dominate the substrate and natural structure or refugia are absent. This reduction in heterogeneity subjects individuals to competition, predation and extirpation (Gause 1932).

Generating Heterogeneity: Construction of Fish Hooks in the Humber River Induces Community Change
Historical data from the LHR watershed documents the presence of 54 fish species; however, only 22 fish species were captured in 2004 aquatic surveys. Monitoring data indicates the LHR watershed only supports generalist, tolerant and cool-warmwater fishes, whereas habitat specialists and higher trophic species are absent from the watershed (TRCA 2008).
Construction of artificial habitat can regenerate heterogeneity in anthropogenically simplified habitats (Gore and Bryant 1988). Stream and river restoration is predicated on generating a diversity of available habitats. Current deflectors, dams, cover structures, and bank protection elements are regularly installed to provide diversity in abiotic conditions (Gore and Bryant 1988). Restoration of heterogeneity in the LHR was performed using standard river restoration practices; however, in 2006 the Toronto Region Conservation Authority (TRCA) implemented a novel restoration method to increase the diversity of habitats available for fishes (TRCA 2008). The TRCA constructed 2 fish hooks at the mouth of the Humber River to supply static, low velocity backwater areas in the Humber River. The fish hooks form 2 semi-enclosed areas along the east bank of the Humber River (Figure 1). The structures deflect and concentrate flows, entrain bedload sediments, encourage the establishment of emergent vegetation, and provide small eddy pools for habitat and primary production. The fish hooks are comprised of 3 layers: a 300-mm thick granular bedding stone, forming the base of the structure against the riverbed; a 600-mm thick riprap stone affixed to the base; and armor stone at the surface interface. The fish hooks are 30 m wide at the base and jut out 8 m into the river.
To determine the fish community response to the novel habitat, we assessed both pre-and post-fish hook communities by electroshock sampling. Between 1989 and 2010 at a site adjacent to the fish hooks (43°37'54.7568", -079°28'18.7388"), we performed boat electrofishing 25 times (6 pre-construction and 19 post-construction). We performed boat electrofishing along a 350-m transect sampled upstream and downstream, executing each complete transect over a period of 1000 s. We subsequently transformed the site-by-species data into a similarity matrix, which we produced with a Pearson's Phi resemblance measure for multivariate presence/absence data was produced. From the dissimilarity matrices, we produced a 2-dimensional PCoA ordination to facilitate visual interpretation, where similarity among sites is concordant with distance in the ordination (Legendre and Legendre 1998).
The ordination indicates 2 clusters of points that align with pre-and post construction sampling (Figure 2). An analysis of similarities (ANOSIM) procedure, which measures the variation within groups relative to the between group variation, indicated the groups were statistically different (ANOSIM R = 0.502, p = 0.001). The results demonstrate that post-construction sampling points are more similar to post-construction points than pre-construction points. The similarity within groups is generated by species differences between groups and suggests the construction of the fish hooks induced the differences.
Following the construction of the fish hooks, we detected the presence of several fish species that had been absent from the LHR over the last decade, including alewife (alosa pseudoharengus), golden shiner (notemigonus crysoleucas), bluntnose minnow (Pimephales notatus), and common shiner (luxilus cornutus). Rainbow smelt (osmerus mordax) and Alaskan stickleback (gasterosteus aculeatus), which were last detected in the LHR in 1989 and 1999, respectively, were also present. We captured most species in low numbers, detecting only 1 golden shiner and Alaskan stickleback, 4 bluntnose minnows, and 6 rainbow smelts. However, alewife and common shiner demonstrated marked increases in abundance in post-construction sampling.
Prior to the installation of the fish hooks neither alewife nor common shiner was detected at the site, but we captured 1,344 alewife and 181 common shiner individuals in postconstruction sampling. In addition, we detected a novel species, a singleton spotfin shiner (cyprinella spiloptera), which had never been recorded in the LHR watershed. The results suggest that construction of the fish hooks create habitat that is suitable for species in higher trophic levels.
We also detected the reoccurrence of 3 piscivorous species, including rainbow trout (oncorhynchus mykiss), largemouth bass (Micropterus salmoides), and yellow perch (Perca flavescens). These species had not been captured in the watershed since at least 1999. Post-construction sampling detected the presence of 1 rainbow trout, 1 largemouth bass, and 3 yellow perch individuals. However, 3 species, American eel (anguilla rostrata), bluegill (lepomis macrochirus), and white perch (Morone americana), that were present during pre-construction sampling were not detected in subsequent sampling.
Our results suggest the installation of fish hooks generate heterogeneity in the fish community. The availability of backwater areas increased the structural heterogeneity of the Humber River and facilitated the persistence of species that are precluded by the velocity of the water. The composition of novel detections was dominated by species with low velocity habitat preferences. Golden shiner is partial to slow moving sections of rivers, whereas bluntnose minnow, rainbow smelt, Alaskan stickleback, and spotfin shiner prefer the pools of larger rivers (Page and Burr 2011). We hypothesize that the detection of novel piscivorous species at the sampling site was a function of an increase in prey availability mediated by the increase in habitat availability. The loss of American eel post-construction could be attributed to its overall widespread decline, and low abundances of bluegill and white perch could be due to a sampling effect, given the species were originally detected as singletons. A description of the niche generated by the fish hooks is required to understand the species patterns observed. A characterization of the abiotic factors of the habitat within the fish hooks would elucidate the relationship between the habitat species composition. Though the mechanism is speculative, the detection of novel, relatively rare species and piscivores indicates a positive response to the construction of the fish hooks. The fish hooks appear to provide additional habitat for species and a resource base for piscivores. These results indicate the construction of fish hooks can be a viable restoration technique to provide novel habitat for deteriorated rivers. Legendre, P. and L. Legendre. 1998. numerical , Nagel et al. 2008. However, as is often the case, plants repeatedly sprayed with the same herbicide can form resistance, eliminating the effectiveness of the selective herbicide (Holt and LeBaron 1990). While resistance by glossy buckthorn to glyphosate has yet to be documented, glyphosate resistance has been documented in several agricultural weeds (Yu et al. 2007). To be proactive, managers at SNWR should consider alternative methods for glossy buckthorn management before glyphosate resistance develops.
Triclopyr (brand name garlon® 4 Ultra, Dow AgroSciences, LLC, Indianapolis, IN, USA) is an herbicide indicated for management of woody plants and herbaceous broadleaf weeds, including glossy buckthorn ( product label). However, little information exists in the literature about the efficacy of foliar applications of triclopyr. Most research conducted with triclopyr on glossy buckthorn and closely related European buckthorn (Rhamnus cathartica) has utilized either a cut-stump or basal bark application at relatively high concentrations (>15%) of active ingredient mixed with mineral oil or diesel oil (Pergams and Norton 2006). However, these methods are less attractive than foliar application because cut stems may resprout vigorously and seedlings too must be managed , Nagel et al. 2008, while the use of fuel oils further increases costs. Moreover, as suggested by Relyea (2005), using the lowest effective concentration is favorable to reduce negative human and environmental health risks. This study tested the efficacy of varying concentrations of triclopyr to determine the best strategy for glossy buckthorn management at SNWR. The study site was located on an anthropogenic dike comprised of coarse sands within SNWR. In June 2011, I set up 40 1-m2 plots comprised of glossy buckthorn resprouts and seedlings with at least 1m between each plot. At each plot, I recorded the number of glossy buckthorn stems, as well as the average stem height and diameter. The mean number of pre-treatment stems (± SE) was 14.8 (± 1.6). The majority of the plots (63%) had stems <1 m in height, with the remaining stems between 1-2 m tall. Mean stem diameter was <2 cm in all but 2 of the plots. I then divided the plots evenly into 4 treatment groups: 0% active ingredient (tap water without surfactant, control), 1.25% active ingredient, 2.5% active ingredient, and 5% active ingredient, following the methods of Corace and colleagues (2008). I mixed each solution using a base concentration of 60.45% active ingredient triclopyr diluted with tap water per label recommendations. On 6 June 2011, I treated the plots with low-volume hand-held sprayers during appropriate weather conditions. I treated each stem to a point of about 50% foliar coverage. I then monitored the plots once per week for 5 wks to document a stress gradient from chlorosis (least severe stress), to shriveled leaves, to shriveled/no leaves and brittle stems (indicating a dead stem). After 2 wks, severe chlorosis was apparent on all treated stems, regardless of concentration. Within 3 wks, 99% of stems were dead, with 100% stem mortality observed in the 4th week ( Figure 1). These results were similar to those of Corace and others (2008).
Our results indicate that foliar applications of triclopyr at low concentrations mixed with water can be effective for managing glossy buckthorn in Upper Michigan, adding to the repertoire of management techniques for this invasive plant species. While complete removal of invasive plant species is highly unlikely, a broad range of management techniques can help to reduce their geographic extent and dominance over several treatment periods (Pergams andNorton 2006, Nagel et al. 2008). Herbicide resistance to glyphosate and triclopyr has yet to be documented in glossy buckthorn. Utilizing a rotation of these 2 herbicides will greatly reduce the likelihood of glossy buckthorn developing a resistance. This is especially important when managers may be limited in techniques available due to constraints such as wetland approved herbicide use or lack of resources. Based on this study and past studies within the SNWR landscape , Nagel et al. 2008), a number of potential research questions could still be addressed that would guide land management actions. First, given more time, managers should address glossy buckthorn regeneration in the triclopyr treated plots in the next growing season to determine resprouting response. Also, if herbicide resistance could be documented in areas that have been treated for several growing seasons with the same herbicide, what is an appropriate model for addressing questions about herbicide rotations? Finally, land managers may want to test competing hypotheses regarding rates of colonization on uninfested areas because these areas could represent spatially distinct subpopulations that may still be effectively managed by glyphosate.