The location of residues is important because surface litter decomposition is generally slower and more variable than residues incorporated into the soil via root deposition and faunal activity. This is because surface litter is more susceptible to extremes in temperature and moisture, nutrient elements that have been mineralized are more vulnerable to runoff or volatization, and surface litter is less accessible to most soil organisms except larger fauna such as earthworms and fungal mycelium (Brady and Weil 2002). Also, if the surface litter is low in nitrogen, fungi may transfer nitrogen from the soil through their hyphae to narrow the C/N ratio of the litter thereby depleting the nitrogen in the soil.
The quantity and quality of organic matter from the over- and understories also changes after thinning. Decomposition thinned stems and detritus in younger forests is higher than stems in mature forests because the sapwood volume is relatively greater in woody detritus from young trees than from old trees (Harmon et al. 1990), and leaf litter decay is greater in younger stands where temperature and moisture conditions are more favorable (Edmonds 1978). That said, the decay rate of any individual piece of dead wood is a function of substrate quality, microbial activity, air temperature, and available moisture (Yin 1999).
12 There are also large differences in biomass production and tissue nutrient concentrations between different tree species, which affects soil properties such as pH, nutrient cycling, and soil biota (Binkley and Giardina 1998). For instance, red alder (Alnus rubra), which is a common early successional species in BC, is a nitrogen fixer. Red alder are associated with Frankia, which are a genus of Actinomycetes that convert inert atmospheric dinitrogen gas to nitrogen-containing organic compounds. This symbiotic relationship lets alder colonize infertile soils and or highly disturbed soils which are often inhospitable to other plant species because of poor nutrient (mainly N) conditions that limit plant growth. Over time, alder builds the nitrogen capital of the soil through leaf litter and root exudates which in turn makes the site more hospitable to other species (Brady and Weil 2002). Furthermore, consistent large effects of nitrogen-fixing trees on soil carbon storage have been documented. However, it is not known if the higher carbon storage is a result of greater carbon inputs or reduced carbon outputs (Resh et al. 2002). 2.6 Soil Carbon Dynamics Globally, soils contain twice as much carbon as vegetation or the atmosphere, and changes in soil carbon content can have significant impacts on the global carbon budget (Bellamy et al. 2005). Soil organic matter (SOM) contains approximately 1500 Pg C to a depth of 1 m (Eswaran et al. 2000). Soil organic matter is composed of accumulated, decaying plant and animal matter on or in the soil; this includes everything from carcasses from recently deceased soil invertebrates to millennia-old humified plant material (Janzen 2006). The principle source of soil organic matter is plant tissue; approximately 42% of dried plant tissue is carbon (Brady and Weil 2002). Soil carbon is

13 part of a dynamic cycle; the carbon content of a soil at any given time is a function of the rates of addition from photosynthetic C plant growth versus the rates of removal from decomposition, leaching, and other soil processes. Soil c sequestration may be managed to help slow the rise of atmospheric CO2. This is just one of SOM benefits though because soil containing more organic matter is more productive and has persistent benefits through its physical effects on soil structure and moisture retention, and chemical effects such as ion exchange (Janzen 2006).
There are three mechanisms by which biomass are transferred to DOM: litterfall, mortality associated with stand breakup in the overmature stand growth phase, and disturbance (Kurz and Apps 1999). Litterfall is composed of all annual transfers of biomass to dead organic matter (DOM) C pools. There are five main pathways through which C enters the soil; from litter, by transfer from roots to mycorrhizal fungi, directly into the soil as mycorrhizodeposits or secreted enzymes, and through grazing by soil fauna (Millard et al. 2007). Notably, the speed at which organic matter moves through these pathways is variable. For example, foliage has a turnover rate of around once per year, whereas fine root turnover occurs about three times per year (Lukac et al. 2003).
Soil fungi in coniferous forests consist mainly of molds and mushroom fungi. Molds are a filamentous fungi that play an important role in soil organic matter decomposition and are widely distributed (Brady and Weil 2002). Mushroom fungi are associates with trees where there is lots of moisture and organic residue and are extremely important in woody tissue decomposition (Brady and Weil 2002). Below ground fungal inputs of
14 biomass are considerable; particularly the fine root and mycorrhizal fungal component of the biomass pool (McDowell et al. 2001). In a study of Douglas-fir forests in western Oregon (Fogel and Hunt 1983), total mycorrhizal and saprotrophic hyphal biomass was estimated to be ca. 660 g m-2. Furthermore, the C in extrametrical mycelium and associated bacteria form a carbon pool with a fast turnover rate (Godbold et al. 2006). The combination of high turnover rates and large biomass associated with mycorrhizal hyphae may prove to be a fundamental mechanism for the transfer of root derived C to soil C (Godbold et al. 2006). Furthermore, both ectomycorrhizal and arbuscular mycorrhizal fungi contain relatively recalcitrant compounds, chitin and glomalin respectively, which remain in the soil following fungal senescence (Treseder et al. 2007). Therefore, an increase in mycorrhizal hyphal biomass should increase C sequestration in forest soils
Rhizodeposition occurs when trees release labile C through the sloughing of root cells, and the release of low molecular mass exudates and organic secretions from their roots to adjacent soil (Phillips and Fahey 2006). This release causes alterations of the physical, chemical, and biological characteristics of the soil around roots and is known as the rhizosphere effect (Phillips and Fahey 2006). For most free living soil microbes, C substrates such as sugars, organic acids and amino acids are limiting factors for growth which explains why generally there is greater microbial activity in the rhizosphere in comparison to bulk soil (Millard et al. 2007). Furthermore, rhizodeposits can also act as primers for the degradation of existing SOM (Millard et al. 2007, Dijkstra and Cheng 2007). Thus an increase in C inputs into the soil in the rhizosphere will not necessarily lead to increased soil C storage, instead, enhanced soil organic carbon deposition may

19 Recently, plantations have become a mechanism for fixing carbon (Brockerhoff et al. 2008). In terms of silviculture practises within plantations, the three primary factors influencing optimum C storage are rotation length, the amount of live mass harvested, and the amount of detritus removed from the forest through slash burning; carbon stores increase as rotation length increases but decrease as the amount of biomass and detritus removed increases (Harmon and Marks 2002). Understanding decomposition processes and the influence of forest management practises on them, and thus the carbon cycle, is crucial to maintaining the long-term productivity and carbon sequestration potential of managed forests. However, there is a lack of strong relationships of carbon pools with individual variables across different ecosystem types. This suggests that there is a complex interplay between climate, species composition, stand age, and soil properties such as texture (Homann et al. 2005). Any restoration plan that aims to maximize carbon sequestration while maintaining biodiversity values must be ecosystem specific.
In terms of total C storage, there is between 2.2 and 2.3 times as much storage in 450 year old natural stand of Douglas-fir-Western Hemlock (Tsuga heterophylla) then in a 60 year old Douglas-fir plantation (Harmon et al. 1990). However, the process of stand biomass development and C accumulation is accelerated in plantations compared to natural stands due to the higher initial density of stems in plantations and subsequent earlier crown closure (Long and Turner 1974). For example, total tree biomass in a 73 year old natural stand is comparable to 42 year old plantation (Long and Turner 1974). Thus stand age and structure play an important role in determining the magnitudes and patterns of C cycling processes within forested ecosystems (Humphreys et al. 2006). In
20 the context of soil carbon, there does not appear to be a trade-off between restoring a coastal Douglas-fir plantation for biodiversity through thinning versus carbon sequestration if the measures used to increase biodiversity do not reduce organic matter accumulation on the forest floor. If the amount of C input into the soil was fixed, decay would need to be suppressed to increase the carbon stores. However, if the restoration measures used on site do not remove any organic matter, then all thinned trees and leaf litter are left on site and add to the detrital component of carbon storage.
Importantly, although forest plantations sequester some carbon while alive, due to the simplified nature of these forests, it is more likely that at some point they will succumb to disease, insect outbreaks and/ or fire and therefore may not be considered reliable carbon offsets (Wilson and Hebda 2008). Alternatively, healthy, functioning, diverse ecosystems tend to be more resilient and therefore less vulnerable. There is the potential to restore forest plantations to a healthy functioning state. The question is, how do we restore a plantation forest so that it will simultaneously optimize carbon sequestration and increase function and biodiversity?

3.2.1 Island Biogeography Theory
According to the island biogeography theory, the number of species in a remnant patch of forest is not only controlled by the habitats and resources present on site, but also by the balance of immigration and local extinction (MacArthur and Wilson 1967). Patterns of immigration are primarily determined by distance from other sources of potential colonists. Habitat patches that are relatively close to other patches are more likely to be occupied then more isolated patches because they are likely to be recolonized after a local extinction event (Pulliam and Johnson 2002). Larger patches can support larger populations, which are less vulnerable to extinction (MacArthur and Wilson 1967). All other things being equal, smaller remnant patches support fewer native species then larger patches (Bellamy et al. 1996). This is not only because larger forest remnants are likely to have a higher ratio of colonisations to extinctions, but also because they are more likely to have undisturbed components necessary to some species (Harris 1984), and are more likely to contain a range of habitats for different species (Fox 1983).
3.2.2 Metapopulation Theory
Metapopulation theory can be thought of as an extension of island biogeography from habitat patches to population patches. Patchy populations are true metapopulations only if movement between subpopulations is neither very common nor very uncommon (Hanski and Simberloff 1997). Clusters of populations may interact over time through the exchange of individuals or genetic material, and individual populations may frequently go extinct and the same area recolonized at a later time by immigrants from extant
25 populations (Pulliam and Johnson 2002). The dynamic nature of local extinctions and recolonization dictate that any particular patch of habitat may or may not be occupied at a given point in time, however, the metapopulation as a whole persists because some patches are always populated (Pulliam and Johnson 2002). In addition, large populations are less likely to go extinct than small populations, and large habitat patches, which are more likely to support large populations, are more likely to be occupied than small patches.
3.2.3 Patch-Matrix-Corridor Model
An extension of the aforementioned fragmentation theories is the patch-matrix-corridor model (Forman 1995). Within large patch and matrix landscapes, disturbances create a diverse, shifting mosaic of successional stages and physical settings of different origin and size (Bormann and Likens 1979). The patch-matrix-corridor model is significant because it recognizes that the ability of a species to reach remnant forest patches depends on how inhospitable, or permeable, the landscape matrix surrounding the patch is (Forman 1995). With this model we move away from the often misleading conceptualization of landscapes as areas of forest/ habitat or non-forest/ non-habitat, toward the idea that the landscape matrix surrounding remnant forest patches may be neither uniformly unsuitable as habitat nor serve as a complete barrier to the dispersal of forest taxa (Kupfer et al. 2006). Thus, the extent to which fragmentation affects a given species depends on how the landscape has been modified, what constitutes suitable habitat for the species, mode and scale of movement, and dispersal behaviour (Fischer and Lindenmayer 2007). Furthermore, the rate of recovery of an ecosystem or species at any scale following a disturbance is not only strongly influenced by the availability of

39 species in the face of climate change (Poiani et al. 2000); however, it has been suggested that the current trend of forest fragmentation may negatively affect the ability of species to undertake large-scale movements such as seasonal migrations and climate changeinduced range shifts (Soule et al. 2004). Furthermore, the effects of climate change on forest biodiversity can act indirectly, cumulatively with other disturbances, or possess feedbacks or time lags.

3.3.3 Invasive Species

Invasive alien species are considered the second largest threat to native species biodiversity next to habitat loss (Wilson 1992). Ecosystem processes are likely to be affected by the invasion of novel organisms (Tilman et al. 1997b). Non-native invasive species are able to alter ecosystem properties if: a) they have considerably different capacities to acquire and use resources than native species, b) they change the trophic organization at the invasion site, or c) they modify disturbance frequency and or intensity (Vitousek 1990). For example, adding novel generalist herbivores to a system can depress producer populations and/or net primary productivity (Vitousek 1990). Even if introduced species do not become invasive, novel species have the potential to alter community structure and ecosystem processes through predation, competition, their potential role as a pathogen, as vectors of diseases, and through their effects on water balance, productivity, and habitat structure (Drake et al. 1989). For example, soil nutrient availability may be diminished through the introduction of plants that produce lowquality acid litter (Vitousek 1990). Furthermore, fire intensity can be heightened in
40 the presence of the invasive species Scotch broom because it is highly flammable due to its high oil content.
3.4 Contributions of Second growth Forests to Biodiversity Biodiversity varies considerably with stand age, but not all species need oldgrowth forests to fulfill their habitat requirements. At different points in successional development, younger stands can provide sufficient structural complexity, species diversity, and ecological function to provide habitat for a variety of different species. Although initial regrowth may be relatively homogenous, heterogeneous forest ecosystems emerge when underlying edaphic and microclimatic gradients (Samuels and Drake 1997), combined with differential success of colonizing species in microsites, eventually cause local divergence in species composition and distribution (Harrelson and Matlack 2006). The precise time scale of compositional divergence is unclear (Harrelson and Matlack 2006); though, as time from disturbance increases, second-growth communities follow a trajectory of increasing richness and changing composition that approaches nearby primary forests (Flinn and Marks 2004). It is often the goal of ecological restoration to speed up this transition. Specific management tools can be applied to second-growth forests to recruit various habitat characteristics, such as wildlife trees, in a much shorter time frame than would occur with natural regeneration alone. Habitat recruitment will be discussed in more detail in Chapter 5.

Wildlife trees are generally created through mortality agents such as fire, disease, insects, windthrow, snowpress, lightning, and wildlife excavation. Death by fire produces a very different type of wildlife tree than gradual death by insects or disease, and tree species and local climate also influence the way a tree deteriorates and decays
50 (Thompson et al. 2005). The most significant indicators of wildlife tree quality are height, diameter, decay stage, location, distribution, and cause of death; but, the value of any particular tree for a given species depends on a variety of attributes, the most important of which are structure, age, condition, abundance, tree species, geographic location, and surrounding habitat features (Thompson et al. 2005). 4.4 Coarse Woody Debris An important structural feature of a wide variety of old-growth forests in North America is a relatively large quantity of downed big logs (Sturtevant et al. 1997). Piece size, species (Harmon et al. 1986), and disturbance history (Spies and Franklin 1988) all affect the speed and type of decay of any CWD piece; while abundance, size, state of decay, and spatial distribution are all factors affecting the use of coarse woody debris (CWD) by wildlife species (Keisker 2000).
CWD is an important nutrient source and growing substrate for numerous species of bacteria, fungi, saprophytic plants, lichens, and mosses that are essential in decay, nitrogen movement, and other nutrient and moisture cycling processes (Thompson et al. 2005). For example, in one study of 6 Biogeoclimatic Subzones in British Columbia, 70% of the 243 plant species recorded grew on CWD, with 23% of those species being restricted to CWD (Song 1997). Other important functions of CWD include carbon storage, erosion control, buffering microclimates suitable for seedling establishment, cover from predators, shelter and access routes for small mammals during heavy snow cover, and contributions to stream stability, complexity, and geomorphology (Thompson et al. 2005).
Coarse woody debris also provides feeding, breeding, and shelter substrates for many invertebrates, small mammals, and amphibians (Thompson et al. 2005). Dupuis et al. (1995) found three to six times more Plethodon vehiculum salamanders in old-growth forests than in younger second-growth forests in coastal B.C.; this was partly attributed to the availability of CWD. Also, Davis (1998) found that the two salamander species observed in coastal B.C. used different decay classes of CWD. This suggests that oldgrowth forests are more likely to provide suitable habitat for both species then singleaged managed forests because old-growth forests generally provide a range of decay classes.

65 Chapter 6: Study Site 6.1 The Study Site District Lot 63 (DL63) is a 65 ha Douglas-fir plantation within the Pebble Beach Nature Reserve on Galiano Island (Figure 2) and is owned by the Galiano Conservancy Association. The land is bound by a section 219 covenant held by the Islands Trust Fund and the Province of British Columbia. Section 219 covenants are voluntary agreements to conserve land or protect features relating to it; they are agreements between private land owners and designated organizations registered on the land title and are legally binding on the future owners of the property. In the context of DL63 this is important because it means that this parcel of land will never be developed.
Figure 2: Galiano Island District Lot 63

66 6.2 Landscape Context

DL63 is at a narrow point of Galiano island and covers approximately 80% of the width of the island at that location. Two of the parcels adjacent to DL63 are regional parks. Given that the Identified Wildlife Management Strategy for the CDF recommends restoring natural conditions and maximizing connectivity, and two of their prioritization tools for creating wildlife habitat areas include protecting communities that could become part of a network and protecting communities adjacent to natural occurrences of other communities, this is important (Pojar et al. 2004). The Pebble Beach Reserve, which is close to the same size as DL63, is classified as an older second growth ecosystem under the sensitive ecosystem inventory. This is important in terms of the permeability of the landscape matrix surrounding DL63; given the location of DL63 in the context of other wildlife habitat patches there are readily available sources of propagules from neighboring patches and DL63 is more likely to be used by native fauna then if it were more isolated from other forested areas. Furthermore, as restoration of DL63 progresses and the composition and structure of the stand more closely resembles that of the Pebble Beach Reserve, it is more likely to be used by a myriad of wildlife elements. Finally, given the forested state of adjacent land parcels, there is relatively low structural contrast between DL63 and neighboring areas indicating that edge effects are minimized. That said, there are several persistent non-native invasive species along paths, a hydro right-ofway, and the road leading to the forest lot. Theories supporting the importance of landscape context can be found in section 3.2.

67 6.3 Climate

The climate of Galiano Island is dominated by the influence of the rainshadow effect of the Olympic and Vancouver Island Mountains and the moderating effects of the ocean. Galiano Island has a cool Mediterranean climate which is characterized by warm dry summers and mild wet winters.

a.) increasing structural diversity through the creation of snags, short term CWD, and root wad structures; b.) growth promotion in existing large trees; c.) canopy gap creation to encourage understory growth; d.) a reduction in ladder fuels; e.) increasing genetic diversity through the release of native Galiano tree seedlings; f.) increasing species diversity through the release of under represented species such as arbutus (Arbutus menziesii); and the release of rare species such as western yew (Texus brevifolia).
Additionally, some of the larger pieces of CWD from the windrows were erected to provide immediate large snag complexity. The large quantities of wood piled in the windrows were redistributed across the forest floor to increase forest floor complexity and nutrient cycling. Furthermore, all downed thinned trees were left on site to contribute to CWD volume. Some under planting with Western redcedar (Thuja plicata) was also done. 6.5 Hypothesis and Predictions
The first objective of this study was to determine if the restoration was successful in terms of changes in structural attributes. Indicators of successful restoration were defined
73 as a significant increase in the size and number of snags, the amount of CWD, understory species richness and abundance, and a significant reduction in the density of planted Douglas-fir or canopy cover. The null hypothesis was that there would be no difference in these structural attributes between treatment and reference areas indicating that the restoration was unsuccessful. However, I predicted that all of these attributes had changed as a result of treatment.
The second objective of this study was to determine how the restoration affected soil carbon and nitrogen at different depths. Because the treatments took place over different seasons and years, soil is highly variable over space and time, and there is a myriad of direct and indirect pathways related to changes in soil carbon and nitrogen, this was a more complicated endeavour. It was predicted that soil carbon and nitrogen would be most affected in the forest floor and the first mineral layer, and any effects would become less pronounced with depth. Predicted mechanisms driving these changes include an increase in Douglas fir needle litter from thinned trees, a change in the quality and quantity of understory litter, an increase in dead tree root biomass below the forest floor as a consequence of thinning, an increase in carbon volitization from the soil associated with soil disturbance related to pulling over trees, and an increase in carbon leaching into the soil from an increase in the quantity of decaying wood and leaf litter on the soil surface. In the scope of this study it was not possible to measure some of these potential mechanisms directly and it was assumed that tree thinning could be used as a proxy for increases in leaf litter and dead root biomass, and species richness for quality of understory litter. The first null hypothesis was that there would be no difference in soil

1. Formulate a test statistic 2. Define the null hypothesis 3. Create a new data set consisting of the original data but randomly re-arranged 4. Calculate the test statistic for the re-arranged data set and compare it to the original true data set value 5. Repeat steps 3 and 4 hundreds of times 6. If the true test statistic is greater then 95% of the random values, then you can reject the null hypothesis at p<0.05 for a one tailed test (for a two tailed test the cutoff should be 97.5%)
For the randomization test on two independent samples in this study I used a t value; instead of just taking the difference between the means, I also divided this value by the standard error of the difference. I used this because it represents the difference relative to
94 the variability of the differences (Howell 2007). However, the difference between the means will lead to the same conclusion. I used 1000 iterations, I did this because an increase in the number of iterations gives us a better estimate of p, but does not increase the likelihood of significance (Palmer 2010). This randomization process results in 1000 random reallocations of the original observations to two groups, and because the data was randomly shuffled into two groups, the distribution of these permutations shows what t values we would be likely to obtain when the null hypothesis is true. The distribution of the permutations can also be used to identify the presence of outliers; outliers would turn up in every permutation, sometimes in one group and sometimes in another, and would cause the resulting t value to swing back and forth between positive and negative, creating an unusual sampling distribution. There were no bimodal sampling distributions in my permutations. I then went back to the original data and asked, would the results that I obtained be likely to arise if the null hypothesis were true? If the probability of obtaining a t values as extreme as the original data was less then =.05, I rejected the null hypothesis and concluded that there was a difference between the groups under consideration.
Correlations were used to assess the presence and strength of relationships between structural variables and soil variables for all soil layers exhibiting significant differences in the previous component of the analysis, namely the forest floor and the first mineral soil layer. The number of samples for this component of the analysis was higher (n=29) than for the previous analysis which used a range of sample sizes depending on the soil layer being analysed. It is expected that noise in the data resulting from year of
95 treatment may impede the ability to detect relationships. However, it is not possible to perform these correlations according to year of treatment because the sample sizes would be much smaller then what is recommended for correlations. That said, scatter plots presented with the results do differentiate year of treatment and can be used as a tool to examine and explain the noise in the data. And because parametric assumptions regarding normality and heterogeneity were violated, Kendalls tau and Spearmans rho were used. Spearmans rank correlation is more widely used then Kendalls tau because it is easier to compute. However, the distribution of Kendalls tau has slightly better statistical properties and there is a direct interpretation of Kendalls tau in terms of probabilities of observing concordant and discordant pairs (Conover 1980). Spearmans rho is interpreted the same way a parametric Pearsons correlation coefficient is, i.e., the proportion of variability accounted for. Kendalls Tau, on the other hand, represents a probability, i.e., the difference between the probability that the observed data are in the same order versus the probability that the data are not in the same order.

9.1.2 Tree analysis and Interpretation
Two species of trees were examined in the analysis, Douglas-fir and red alder. The dominant tree in the stand was Douglas-fir which represented 67% of the trees in the treatment area and 77% of the trees in the reference area. The sub-dominant tree was red alder which represented 23% of the trees in the treatment area and 9% of the trees in the reference area. All other tree species combined represented only 10% of the trees in the treatment area and 12% of the trees in the reference area and thus were considered negligible. These percentages were calculated separately for treatment and reference areas, the only tree species transferred from the living biomass pool to the dead biomass pool was Douglas-fir, so changes in proportions are to be expected. Another reason why
102 red alder has been included in this analysis is its well documented influence on soil nitrogen and carbon pools. This will be discussed in detail in the soil analysis and interpretation section. However, from a structural standpoint it is important to note that there was a significant difference in Douglas-fir volume (z = -2.803, p = 0.004) and density (z = -3.598, p = 0.000) between treatment and reference areas (Figures 7 and 8 respectively), but there was no significant difference in red alder volume (z = -0.393, p = 0.709) or density (z = -0.474, p = 0.672) between treatment and reference areas. In other words, there were significantly fewer Douglas-fir in the treatment area then in the reference area, and approximately the same number of red alder in the treatment and reference areas.
Figure 7: Reference vs. treatment Douglas-fir volume, points in figure identified by number refer to outliers
Figure 8: Reference vs. treatment Douglas-fir density
9.1.3 Coarse Woody Debris Analysis and Interpretation
In terms of coarse woody debris, both the number of fragments (z = 3.628, p = 0.000) and volume (z = 2.319, p = 0.020) was significantly higher in the treatment area then in the reference area (Figures 9 and 10 respectively). The mean number of pieces across treatment area was 9909 per hectare, whereas the mean number of pieces across the reference area was 2285 per hectare. The mean volume across the treatment area was 954.55 m3/ha where as the mean volume across the reference area was 288.57 m3/ha. The interpretation of this difference is that the restoration treatments resulted in a significant increase in the amount of CWD in the treatment area.

Figure 25: Correlation between forest floor soil carbon and nitrogen
Figure 26: Correlation between forest floor soil nitrogen and pH
9.3.2 Mineral Soil Layer 1
There were only two statistically significant relationships found within this component of the analysis. There appears to be a weak negative relationship (rho = -0.395, p = 0.034) between alder volume and mineral soil layer 1 pH, and a strong relationship (rho = 0.738, p = 0.000) between mineral soil layer 1 carbon and mineral soil layer 1 nitrogen. Figures 27 and 28 depict these relationships and will be discussed in more detail in chapter 10.
Figure 27: Correlation between mineral soil layer 1 pH and red alder volume
Figure 28: Correlation between mineral soil layer 1 carbon and nitrogen

10. Discussion

It is the goal of ecological restoration to emulate the natural composition, structure, and functioning of a degraded ecosystem in a manner that will leave it self sustaining and integrated with the ecological landscape in which it occurs (Higgs 1997). It is apparent that the restoration treatments have changed the structure and composition of the stand and that the treatments have influenced nutrient cycling processes within the forest floor and the mineral soil to a depth of 15 cm.
The restoration practices applied to DL63 generally increased the biodiversity values within the stand. Both Douglas-fir and released volunteer species such as red alder and big leaf maple provide resources and structures favourable to native flora and fauna; the implementation of wider tree spacing and canopy gaps increased understory vegetation; and the recruitment of structural attributes such as snags and CWD resulted in a significant increase of said attributes throughout the stand (Section 4.6). Following is a brief discussion of the most important measured variables highlighting the influence of the restoration.

10.1 Structure

10.1.1 CWD
The CWD that has been moved from the windrows to the forest floor seem to have been placed in a way that mimics its natural distribution of randomness and connectivity, with
127 some clumping and layering. These pieces could contribute to the structural diversity of the forest floor for quite some time providing they are of sufficient size, species, and decay stage. However, it was not possible to determine species for most of the CWD fragments due to there advanced decay status; very few CWD pieces pulled out of the windrows still had bark on them or other simple species identifying features. Along the same vein, the vast majority of the CWD was in the moderate to advanced decay classes three and four; the implication of this state of decay is a reduced longevity for these pieces. In other words, these pieces will likely not contribute to wildlife habitat for very long. On the other hand, because the majority of these CWD pieces are in full contact with the forest floor, have soft blocky pieces, and likely invading roots in the heartwood, leaching processes and nutrient cycling from the wood to the forest floor is likely much higher then if they were not as decomposed. Thus the positive relationship between the number of CWD fragments and soil carbon is not surprising. However, the observed relationship may be less direct. It is possible that when the wood was in the windrows it was colonized by fungi, which has potentially entered the soil and is increasing soil carbon.

It was originally thought that thinned trees that had been placed on the forest floor were less significant due to the small size of this wood and its associated rapid decay rates. In other words, it was thought that pulled over trees would only contribute to forest floor structure for a short period of time. However, those thinned trees that are suspended above the ground, like many of the pulled over trees, will decay at slower rates than those on the ground (Mattson et al. 1987). Furthermore, a study done by Edmonds and Eglitis
128 (1989) suggests that it would take approximately 115 years for 24 cm diameter Douglasfir log to reach 95% decay where as it would take approximately 60 years for a 37 cm diameter Doulas-fir log to reach 95% decay. The reason for this seemingly illogical conclusion was that small logs were not as easily attacked by wood-boring insects which actually spread wood-rotting fungus. Thus there is the potential that pulled over trees will be part of the stand for quite some time.
This means that windrow derived CWD can provide important short-term ecological benefits whereas newly created CWD provides ecological benefits for a greater period of time. As of right now, the majority of the CWD pieces are relatively small. Ideally, larger CWD pieces will be recruited as the stand continues to age, which is likely due to the predicted increase in incremental tree growth resulting from thinning treatments. Larger pieces are of higher value than what is currently in the stand because they generally decay more slowly, hold more moisture, present less of a fire hazard, and provide more habitat value to a greater number of wildlife species (Manning et al. 2006).

10.1.2 Snags

Mortality rates are generally highest in young seral stages (Franklin et al. 1987), due to canopy closure and stem exclusion. High mortality rates are evident in DL63 where there are high densities of very small snags irrespective of treatment. However, it is predicted that snag production rates and density will generally fall with increasing stand age, but mean snag size and longevity will generally increase (Cline et al. 1980).
129 In terms of snag recruitment, it is apparent that there are more Douglas-fir snags in the treated areas then the control areas. However, the value of this increase must be discussed in the context of both wildlife use and nutrient cycling. Characteristics that affect the value of individual snags as habitat include cause of death, diameter, tree form, bark condition, tree species, and tree height (Lofroth 1998).
Cause of death is important because the ability of decay organisms such as fungi and insects to invade wildlife trees affects the ability of other wildlife elements to use said trees. For example, decay organisms further weaken or soften tree tissue which allows primary cavity nesters to excavate nests (Thomas et al. 1979). This is evident in DL63, where only the topped trees and erected snags showed any evidence of wildlife use, generally in the form of woodpecker excavations. This is likely due to the fact that topping a tree more closely mimics natural processes then girdling a tree; when the top of a tree is broken off it provides a location for water to pool and spores to land which may result in fungal rot from the inside out. Girdling a tree on the other hand, may disrupt the trees respiration process and ultimately kill the tree, however, where the tree is wounded it will exude sap effectively preventing fungi from entering the tree. The result is girdled trees likely rot from the outside -in vs. the inside-out which may have implications for wildlife use of said trees. The work done by Todd Manning with fungal inoculation may be a more effective alternative to girdling trees.

ShapiroWilk 0.012 0.107 0.481 0.107 0.122 0.075 0.127 0.121 0.001 0.056 0.081 0.046 0.113 0.017 0.0.084 0.054 0.491 0.002 0.163 0.795 0.414
Transformation SQRT SQRT SQRT SQRT LN LN LOG 10 NA LOG 10 SQRT SQRT NA NA LOG 10 LOG 10 NA NA LN, SQRT LN, SQRT NA NA LN LN
ShapiroWilk 0.043 0.376 0.043 0.079 0.079 0.305 0.264 0.663 0.573 0.783
Levene's Test 0.092 0.545

0.369 0.945

0.605 0.394

.113,.122.696,.604

0.03 0.33

0.142 0.392

155 Appendix B Structural Analysis Results Summary

Variable

Mann-Whitney

Kolmogorov-Smirnov

GLM Univariate Analysis Sig. 0.335; Observed Power 0.157 NA Sig. 0.068; Observed Power 0.451 Sig. 0.010; Observed Power 0.764 Sig. 0.005; Observed Power 0.828 Sig. 0.003; Observed Power 0.875 NA NA NA Sig. 0.000; Observed Power 1.000 NA Sig. 0.000; Observed Power 1.000
Natural Snag Density Natural Snag Mean DBH Total Snag Density

0.409 0.901

0.779 0.958
T-test Equal Variance Assumed 0.335, Equal Variance Not Assumed 0.487 NA Equal Variance Assumed 0.068, Equal Variance Not Assumed 0.121 Equal Variance Assumed 0.010, Equal Variance Not Assumed 0.038 Equal Variance Assumed 0.005, Equal Variance Not Assumed 0.005 Equal Variance Assumed 0.003, Equal Variance Not Assumed 0.000 NA NA NA Equal Variance Assumed 0.000, Equal Variance Not Assumed 0.001 NA Equal Variance Assumed 0.000, Equal Variance Not Assumed 0.001
0.088 Douglas-fir Snag Volume Natural vs. Created Snag Mean DBH Residual Douglas-fir Volume Residual Douglas-fir Density Red Alder Volume Red Alder Density

0.004 0.000 0.709 0.672

0.004 0.004 0.997 0.995
CWD # of Fragments CWD Volume

0.000 0.020

0.001 0.141

Species Richness

156 Appendix C Treatment Year and Reference Site Soil Analysis Summary Soil Variable Soil Layer Treatment Years All 2004, 2006 Ref, 2006 All 2004, 2005 2004, 2006 Ref, 2005 Ref, 2006 All All All Ref, 2006 All 2004, 2005 2004, 2006 All All All All 2004, 2005 2004, 2006 Ref, 2005 Ref, 2006 All All All All Ref, 2004 All Ref, 2004 2004, 2005 All All GLM Analysis* 0.048, 0.639 0.039, NA 0.013, NA 0.006, 0.879 0.021, NA 0.705, 0.127 0.513, 0.154 0.403, 0.242 0.011, 0.829 0.054, NA 0.044, NA 0.011, 0.829 0.207, 0.365 0.558, 0.141 0.001, 0.981 0.014, NA 0.012, NA 0.442, 0.224 0.288, 0.299 0.359, 0.213 0.030, 0.71 0.047, NA 0.023, 0.744 0.042, NA 0.149, 0.430 0.677, 0.113 Randomization** NA 0.008 0.002 NA 0.011 0.016 0.007 0.045 NA NA NA 0.003 NA 0.019 0.013 NA NA NA NA 0.001 0.014 0.001 0.002 NA NA NA NA 0.021 NA 0.009 0.034 NA NA