AN ASSESSMENT OF ECOLOGICAL THINNING FOR STRUCTURAL DIVERSITY IN COASTAL FORESTS OF THE PACIFIC NORTHWEST, USA Ansel Bubel Advisor: Sari Palmroth Submitted in partial fulfillment of the Masters of Environmental Management and Masters of Forestry Degrees from Duke University 12/2013 Abstract Restoration for increased structural integrity is a relatively new strategy that is being applied to protected areas in the Pacific Northwest. Not enough data has been collected to determine how best to approach restoration, so the current projects have made educated guesses about what the best approach might be to increase structural diversity. My analysis seeks to determine the most effective thinning regime to increase structural diversity. The current stands which are undergoing restoration are young even aged plantations which have been acquired by conservation organizations and government agencies. Ecologically, these forests differ from the historical forests both in structure and species composition. Restoration seeks to remove a portion of the overstory trees to allow the recruitment of a new cohort of trees and increase the growth rate of the remaining overstory trees. This approach essentially involves tunneling through the classic development curve laid out by Odum (Odum, 1969). I used a version of FVS ported into the R language to model the background forest growth in the face of windthrow against 25 possible thinning types. I compared thinnings across a range of intensity of tree removal (10-50% of trees removed per cycle) and cycle length (10 to 50 years). I compared these results to a background scenario consisting of windthrow disturbance with a 1% probability of occurring each year and lognormally distributed impacts with low intensity damage more common than high intensity damage. My work has shown that removing 30 – 50 % of basal area every 30 to 50 years can help forests achieve structural complexity before it would develop on its own accord. Without restoration, natural forests achieve structure similar to old growth between 100 and 200 years of age depending on the frequency and intensity of windthrow. After 100 years of restoration, forests will achieve higher levels of structural complexity and tree growth. Table of Contents Abstract ......................................................................................................................................................... 2 Introduction .................................................................................................................................................. 8 Background ............................................................................................................................................... 8 Current Forest condition ........................................................................................................................... 8 Desired Forest Condition .......................................................................................................................... 8 Restoration Approach and implementation ............................................................................................. 9 Project Goal............................................................................................................................................... 9 Methods ...................................................................................................................................................... 10 Modeling Forest Growth ......................................................................................................................... 10 FVS Modeling in R ............................................................................................................................... 10 Baseline forest ........................................................................................................................................ 10 Model Parameters .............................................................................................................................. 12 Results ......................................................................................................................................................... 13 Basic FVS Modeling ................................................................................................................................. 13 Baseline No-Restoration Scenario ...................................................................................................... 14 Development of Treated stands ............................................................................................................. 16 Basal Area............................................................................................................................................ 16 Maximum Stem Diameter ................................................................................................................... 17 Standard Deviation of Stem Diameter ................................................................................................ 18 Discussion.................................................................................................................................................... 19 Development of structural diversity ................................................................................................... 19 Next Steps ............................................................................................................................................... 24 Thinning across a landscape ............................................................................................................... 25 Windthrow .......................................................................................................................................... 26 Conclusion ................................................................................................................................................... 26 References .................................................................................................................................................. 27 Appendix A: Historical Analysis ................................................................................................................... 28 Methods .................................................................................................................................................. 28 Historical Analysis ............................................................................................................................... 28 Historical Results ..................................................................................................................................... 29 Historical Maps ................................................................................................................................... 29 Specific Land Cover Types ................................................................................................................... 30 Weather .............................................................................................................................................. 32 Native uses of natural resources ........................................................................................................ 33 Native trading ..................................................................................................................................... 35 Lessons from the past ............................................................................................................................. 35 Appendix B: Carbon Sequestration ............................................................................................................. 39 Carbon Literature Review ....................................................................................................................... 39 The data problem................................................................................................................................ 39 The classical understanding of forest carbon dynamics ..................................................................... 39 Recent revisions to old growth forests role in carbon storage........................................................... 40 Old Growth Forests, Young Forests and Forest restoration ............................................................... 40 Modeling carbon storage under forest restoration ............................................................................ 43 Carbon Storage ....................................................................................................................................... 44 FVS Model Outputs ............................................................................................................................. 44 Carbon Storage ....................................................................................................................................... 49 Next Steps ........................................................................................................................................... 49 Figures Figure 1: Diagram of rFVS Model developed in the R language around the existing FVS algorithm developed by the USDA Forest Service. This model is used to predict the growth of the stand in the presence of windthrow as a baseline for how stands will grow without treatment. ................................. 11 Figure 2: Diagram of rFVS Model developed in the R language around the existing FVS algorithm developed by the USDA Forest Service. This model is used to predict the growth of the stand for 25 restoration scenarios. The input files used by rFVS were created in the graphical user interface of FVS and imported into R, so the stand setup variables were already called when the file was run in R.......... 11 Figure 3: Potential probability distribution for the severity of wind disturbances for the forests of coastal Oregon. A log-normal distribution was used where light disturbances (0-30% mortality) were more likely than heavy disturbances (70-100% mortality)............................................................................................ 12 Figure 4: Forest at age 45 before treatment has occurred. Notice the even-aged distribution of both the height and DBH. The treatment on this forest removed 20% of trees every 30 years. ............................ 13 Figure 5: Forest at age 140 after the full treatment has occurred. Notice the there is a greater distribution of diameters with the largest trees having diameters greater than 32” compared to the forest before treatment began. This scenario is represents the type of treatment being pursued by both TNC and NPS, though this is not designed to show the exact treatment type that they have implemented. .............................................................................................................................................. 14 Figure 6: Stand variables for the background treatment which experiences log-normally distributed winthrow. [A] shows the trend in basal area and data from stands at Lewis and Clark National Historical Park for a comparison, [B] shows the standard deviation of all trees with diameters greater than 2” DBH, [C] shows the diameter of the largest stem in the plot, while [D] show the percentage of stands which have been affected by windthrow. ............................................................................................................. 15 Figure 7: Restoration scenarios compared with the stand average and bootstrapped confidence intervals from the log-normal background scenario. Note restoration efforts and windthrow begin at age 40, reflecting the stands undergoing treatment at Lewis and Clark National Historical Park and other protected areas. Restoration efforts continue for 100 years, until the stands have reached an age of 140 years old ...................................................................................................................................................... 16 Figure 8: Diameter of the largest stem over the course of stand growth and treatment. Restoration begins in year 40 and continue for the next 100 years to age 140. The average diameter line and the confidence intervals are from the log-normal background scenario for comparison. ............................... 17 Figure 9: Standard deviation of stem diameter over the course of stand growth and treatment. Restoration begins in year 40 and continues for the next 100 years to age 140. The average diameter line and the confidence intervals are from the log-normal background scenario for comparison............ 18 Figure 10: Basal Area after 100 years of treatment versus the annual mortality rate due to thinning. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario. .......................................................... 21 Figure 11: Maximum diameter at age 140 compared with the amount of trees removed per year. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario. .......................................................... 22 Figure 12: Maximum diameter at age 140 for scenarios with an annual removal intensity of 1% as a function of the percentage of trees removed per treatment. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario.................................................................................................................... 23 Figure 13: Standard Deviation of Diameter at age 140 compared to as a function of the annualized thinning removal. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario. ............................. 24 Figure 14: Reconstructed Landuse directly adjacent to Fort Clatsop. Notice the large amount of land accretion beyond Point Adams. This reconfiguration of the mouth of the Columbia has significantly protected the lower river from ocean swells. ............................................................................................ 37 Figure 15: Lewis and Clark’s Description of the landscape of the Lower Columbia River in 1805-1806. Locations are approximate, and in some cases, may refer to whole landscape features beyond the point on the map. ................................................................................................................................................. 38 Figure 16: Carbon storage across ecoregions of the western US, notice that the pacific forests, mainly composed of Sitka Spruce and Western hemlock, store the most carbon, and contain the greatest difference in storage between young and old forests. ............................................................................... 45 Figure 17: A comparison of carbon storage in young and old growth Sitka Spruce Forests compiled from the literature cited in table 3 and 4 ............................................................................................................ 46 Figure 18: Thinning intensity (percent of trees removed per year) versus carbon storage at age 150. Notice that carbon storage is sensitive to low thinning intensities but ending carbon storage values hit a floor after 2% removal per year. This is likely because heavy thinning is followed by more rapid growth due to higher light levels and decreased competition for remaining trees ............................................... 47 Figure 19: Carbon storage versus the proportion of trees cut per treatment. The colored lines correspond to the frequency of thinning treatments. Notice that thinning every 30 – 50 years results in the same ending carbon storage. ............................................................................................................... 47 Figure 20: Predicted Carbon storage for a moderate thinning (15% removal every 30 years) compared with the carbon storage in a young and Sitka Spruce old forests .............................................................. 48 Figure 21: Carbon storage under a moderate restoration thinning compared to a no-management and a timber management scenario..................................................................................................................... 48 Tables Table 1: Suggested treatments (characterized by proportion of trees cut per treatment and frequency of treatment in units of 1/years) for achieving the best stand structure after 100 years indicated in blue highlighted cells of table below. ................................................................................................................. 24 Table 2: Thinning scenarios evaluated to determine the resulting amount of carbon stored in tons/acre. The columns correspond to the proportion of trees cut at every thinning, and the rows denote the return interval of the thinning. ................................................................................................................... 41 Table 3: Comparison of published rates of change for carbon stored in old growth forests ..................... 41 Table 4: Comparison of published estimates of carbon pools in old growth forests ................................. 42 Table 5 Comparison of published estimates of carbon pools young forests .............................................. 43 Introduction Background Several efforts are underway in the Pacific Northwest to improve of the structure of young forests in protected areas. Redwoods National Park began a small pilot restoration project in 1978 and recently expanded their focus given the success of their initial treatments (R. N. Park, 2008). The Nature Conservancy began a restoration project at its Ellsworth Creek in 2008 (Liane Davis, 2009). Lewis and Clark National Park initiated a restoration effort on their forest land in 2012. These efforts all address young stands established by timber companies and then bought by conservation focused organizations. All three projects seek to remove trees from existing stands to reduce competition, thereby promoting the growth of understory vegetation, young trees, and the remaining overstory trees. The initial restoration project at Redwoods National Park has shown that this approach does improve forest structure (R. N. Park, 2008). Yet there have not been predictive studies to determine the most effective timing and intensity of tree removals. Current Forest condition Most of the coastal forests in the Pacific Northwest have been managed for timber production. Managed stands are usually composed of douglas fir (Pseudotsuga menziesii) and western hemlock (Tsuga heterophylla). Stands are planted and managed as a single age cohort and as such, all trees are nearly the same height and diameter. Such forests have little or no understory or young trees. Windthrow is a common source of mortality for both planted and natural stands in the coastal range of Oregon. Commercial sites which experience windthrow damage are commonly logged to recover the downed timber (salvage logging) (C. Clatterbuck, Pers. Comm). Such windthrow has affected some of the organizations with forest restoration programs. At Fort Clatsop, more than 100 acres of this forest was flattened in a winter storm in 2007 (C. Clatterbuck, Pers, Com). Forests managed for timber production are not particularly valuable ecologically because they do not have enough forage to support ungulate browsing and they lack the large trees necessary for certain rare birds. The Northern Spotted Owl (Strix occidentalis caurina) and Marbled Murrlet (Brachyramphus marmoratus) require large old trees to nest in, murrlets living in branches infested with dwarf mistletoe, while spotted owls nest in cavities and holes in tree boles (Forsman, Meslow, & Wight, 1984; Hamer & Nelson, 1995). Young forests in the Pacific Northwest do not provide habitat for these threatened organisms. Desired Forest Condition The organizations which are running forest restoration programs have character as the desired state of the forest in the future. Historical provide a partial record of the original forests which were found across Clark wrote that sitka spruce was commonly 7-12’ in diameter at breast was 4-6’ in diameter [January 8th, 1806, Journals of Lewis and Clark trees attained sizes of 14’ in diameter and 300’ tall. Historical stands show forests of similar proportions. Such stands represent an old maturation was not significantly shaped by humans with the exception somewhat rare along the low lands near the Columbia River due to native use (See Appendix A: Historical Analysis). While the current forests would eventually acquire an old growth character, there is a desire to expedite this transition. Specifically, the park would like to see increased structural diversity, species diversity, understory cover and soil thickness (L. a. C. N. H. Park, 2011). There is not a specific definition of the desired condition outlined in the planning documents released on the various restoration projects. Research into old growth forests has suggested that there are a host of characters which are indicative of old growth forests in the Pacific Northwest. These features include nurse logs, dead snags, large coarse woody debris, abundant understory vegetation, large trees, and relatively high species diversity. These characteristics are not easily predicted using existing models. For the present study, we use basal area (the cross sectional area of tree trunks at 4.5’ from the ground per acre), standard deviation of tree diameter, and maximum diameter as indicators of the forest maturity and diversity. Basal area is a measure of stand density, and while it is not an indicator of old growth forests per se, it is an important measure of stand character. The presence of many different age cohorts of trees is best measured as the standard deviation of diameter. Large trees are important both for habitat, and as a source of snags, coarse woody debris and nurse logs, therefore tracking the maximum diameter will act as a proxy for the presence of these ecologically important aspects of mature forests (M. E. Harmon et al.). Restoration Approach and implementation At Lewis and Clark National Historical Park, the current management plan calls for thinning the young forests (15 to 45 years old) in the park (L. a. C. N. H. Park, 2011). The three projects which are currently underway (Lewis and Clark, and TNC’s Ellsworth Creek Preserve, and Redwoods National Park) use different approaches. Lewis and Clark National Historic Park uses light intensity thinning, which removes 20% of the existing trees every 20 years for 40 or 60 years (Liane Davis, 2009). The Nature Conservancy is removing 50 to 70% of existing trees; however they are only applying this treatment once. These techniques are also planned for implementation on lands owned by other towns along the northern Oregon coast (The town of Cannon Beach). One important thing to note about this approach is that forest restoration in Sitka Spruce dominated forests is new and none of the existing projects are old enough to yield results. We do not know if the current approaches being undertaken lead to an optimal result or whether a different approach might yield better results. Project Goal The goal of the present study is to determine (1) does restoration increase structural diversity compared to a no-management scenario, and (2) what intensity and frequency of thinning produces most structural diversity after 100 years of treatment? Methods Modeling Forest Growth Determining the optimal approach to restoration without a comparison site which has undergone a complete restoration will necessarily require predicting future forest growth. This has traditionally been accomplished with growth and yield tables, however modern computing has allowed these methods to become standardized and made more powerful. Current modeling approaches fall into two categories: Empirical models based on regression equations of actual forests growth (e.g., FVS (Crookston & Dixon, 2005)), and process based models which use physical factors like insolation, crown shape, atmospheric gas concentration and similar measures to determine tree growth (e.g., 3PG)(Landsberg, Waring, & Coops, 2003). While process models offer detailed predictions of certain forest types and processes, parameterizing these models in new forests requires a large amount of data and fieldwork. For this project, processes based models are not a reasonable tool because their basic data requirements are not met by any of the tree inventories conducted at Lewis and Clark National Historical Park. In contrast, FVS has variants for every forest type of the continental US, specific growth equations for all common species, and calibration for local conditions based on FIA data (Crookston & Dixon, 2005). FVS Modeling in R The forest service has developed an R based variant of the FVS model. While intricate scripting is required, the added power of looping and other coding constructs within R expands the capabilities of FVS far beyond what is possible in the standard FVS graphical user interface. For example, random disturbances like windthrow or fire, cannot be addressed by the standard FVS interface, but can be addressed via scripting within R. Within rFVS, it is it easier to make a series of assumptions upfront and run for the entire time period than to let assumptions vary over time. In the rFVS framework, forestry operations are specified in advance and then the core FVS algorithm makes predictions about the stand development(Crookston & Dixon, 2005). It would be possible to modify the source code (built within Fortran) to add time varying characteristics such as windthrow probability and effects. I used a set of simple assumptions and let the model run without varying the basic parameters overtime. Baseline forest The transition to a mature or old growth condition is complex and there is much uncertainty surrounding the species present and the role of disturbance. We know that disturbance, particularly windthrow, is very important for the development of coastal conifer forests in the Pacific Northwest (Nowacki & Kramer, 1998; Taylor, 1990). Yet windthrow is a complex phenomenon which varies in intensity per area, patch size, overall impact, and timing. Therefore, I added windthrow stochastically to the background treatment. The annual probability of disturbance was 1% (Taylor, 1990) and the damage caused by a given event was drawn from a log-normal distribution where removing 10-30% of basal area is three times more likely than removing 70-100% of the stand basal area. The background scenario was run with a 5 year time step and windthrow was allowed to not occur or occur once within the time step. Over the course of the simulation, windthrow was allowed occur in the same stand multiple times. Figure 1: Diagram of rFVS Model developed in the R language around the existing FVS algorithm developed by the USDA Forest Service. This model is used to predict the growth of the stand in the presence of windthrow as a baseline for how stands will grow without treatment. Figure 2: Diagram of rFVS Model developed in the R language around the existing FVS algorithm developed by the USDA Forest Service. This model is used to predict the growth of the stand for 25 restoration scenarios. The input files used by rFVS were created in the graphical user interface of FVS and imported into R, so the stand setup variables were already called when the file was run in R. Figure 3: Potential probability distribution for the severity of wind disturbances for the forests of coastal Oregon. A lognormal distribution was used where light disturbances (0-30% mortality) were more likely than heavy disturbances (70-100% mortality). Model Parameters I sought to model the effect of treating a 40 year old forest in Oregon or Washington. With FVS, I started with a bare stand and planted 600 douglas fir and 600 western hemlock per acre. A precommercial thin was added to FVS at age 10 (to match the most likely historical management on stands undergoing treatment). The treatment period began at age 40 (approximating the present state in 2013), and treatments were applied for the following 100 years. 25 model scenarios were tested and compared with a no-management alternative. These 25 scenarios were all permutations of thinning frequency (every 10, 20, 30, 40 or 50 years) with removal rates (10, 20, 30, 40, or 50 percent of the trees removed in a given treatment). Within FVS, the maximum basal area was set at 300 ft2/ac , as an upper ecological limit for forests in northwest Oregon (based on data from Lewis and Clark National historic Park). The background scenario received windthrow at a 1% probability per year. Windthrow damage was modeled as a lognormal distribution (Figure 3) with the less damaging windthrow (Loss of individual canopy trees) occurring more often than highly damaging windthrow (total loss of a stand). Results Basic FVS Modeling Over the course of a typical treatment, stand age structure changes from an even aged cohort of trees to a multi-aged forest. The 40 year old stand before treatment contains an even aged cohort of western hemlock and douglas fir (Figure 4). Treatment removes a portion of the original cohort of trees and thereby makes room for young trees and understory vegetation to develop (Figure 5). The even aged canopy evident in the younger forest is transformed into a multilayered forest after 100 years of treatment. Figure 4: Forest at age 45 before treatment has occurred. Notice the even-aged distribution of both the height and DBH. The treatment on this forest removed 20% of trees every 30 years. Figure 5: Forest at age 140 after the full treatment has occurred. Notice the there is a greater distribution of diameters with the largest trees having diameters greater than 32” compared to the forest before treatment began. This scenario is represents the type of treatment being pursued by both TNC and NPS, though this is not designed to show the exact treatment type that they have implemented. Baseline No-Restoration Scenario A model of the effect of windthrow, without restoration intervention, on the even-aged forests currently on much of Lewis and Clark Historical National Park provides a baseline for judging the effectiveness of various restoration approaches. The background no management scenarios were tested in FVS and the basal area, trees per acre, and stand volume modeled through the development of the stand (Figure 6). Over the next 100 years, windthrow is predicted to reduce basal area from 300 to 250 ft2/ac, increase the standard deviation of diameter from 5 in to 10 inches, and increase the maximum stem diameter from 20 to 40 inches (see Figure 6). By the end of the scenario, 60% of natural stands are predicted to have experienced windthrow (Figure 6). A B C D Figure 6: Stand variables for the background treatment which experiences log-normally distributed winthrow. [A] shows the trend in basal area and data from stands at Lewis and Clark National Historical Park for a comparison, [B] shows the standard deviation of all trees with diameters greater than 2” DBH, [C] shows the diameter of the largest stem in the plot, while [D] show the percentage of stands which have been affected by windthrow. Development of Treated stands Basal Area Figure 7: Restoration scenarios compared with the stand average and bootstrapped confidence intervals from the lognormal background scenario. Note restoration efforts and windthrow begin at age 40, reflecting the stands undergoing treatment at Lewis and Clark National Historical Park and other protected areas. Restoration efforts continue for 100 years, until the stands have reached an age of 140 years old Stand basal area is a measure of tree density and an important factor in stand development. The results of modelling forest growth in FVS show that both the background average and treatment basal areas decrease modestly through time as shown by the downward trend of most lines in Figure 7. Removals due to disturbances or treatments remove living trees and thereby decrease basal area. Basal area then rebounds as the remaining trees increase their radial growth, and a new cohort of woody vegetation forms. Basal areas for some treatments reaches a low of 50 ft2/ac, which means that treatments are removing too many trees and tending to push the stand towards a more open, early successional state. Maximum Stem Diameter Figure 8: Diameter of the largest stem over the course of stand growth and treatment. Restoration begins in year 40 and continue for the next 100 years to age 140. The average diameter line and the confidence intervals are from the log-normal background scenario for comparison. Large trees are often hallmarks of old growth forests in the Pacific Northwest, and therefore the size of the largest stem in a plot is a measure of stand maturity. Thinning treatments are predicted to increase the maximum diameter above the no-restoration baseline (“background average” in Figure 8). Thinning removes competing vegetation and allows the remaining vegetation to grow faster and increase diameter growth. Natural disturbance has the same effect of removing competing vegetation, however natural disturbance is random and after 100 years, only 60% of the stand has experienced windthrow. Therefore, thinning increases diameter growth like natural disturbances, but because restoration thinning occurs earlier, trees on average grow bigger with treatment. Standard Deviation of Stem Diameter Figure 9: Standard deviation of stem diameter over the course of stand growth and treatment. Restoration begins in year 40 and continues for the next 100 years to age 140. The average diameter line and the confidence intervals are from the lognormal background scenario for comparison. Standard deviation of stem diameter is a measure of how many tree size classes are present in a stand. This metric is a proxy for the complexity of forest structure. Predictions for all treatment scenarios show increasing standard deviations through time, though certain scenarios have higher standard deviations than others. However, all treatment scenarios have predicted stem diameter standard deviation at or below that of the baseline no–restoration scenario. This suggests that no treatment increases stand complexity above the background treatments, and many decrease stand complexity relative to the background. Discussion Development of structural diversity Thinning treatments have been implemented with the goal of increasing the structural diversity above the natural background. The results from rFVS show that treatments increase some metrics of structural complexity while decreasing others. Basal area decreases for all scenarios, but for some scenarios it reaches values well below the background scenario. In these scenarios, treatment removes trees faster than they can grow back until the stand regresses to a point with a higher growth rate. All treatments appear to increase the maximum stand diameter above what would expected in the background case. This is likely because thinning removes competing vegetation and gives the existing vegetation more space, which results in higher growth rates. This is specifically a result of introducing tree removals earlier in the treatments compared to the background. Standard deviation points to a counter point, where treatments have lower standard deviations than the background scenario, suggesting that stand diversity is actually decreased. However, the standard deviation of diameter might be smaller if there are many more young stems compared to the mature canopy stems. Therefore, the decrease in standard deviation (of stems greater than 2” DBH) might reflect the regrowth of young trees rather than a reduction of structural complexity. Standard deviation is calculated on a per tree basis and recruitment of numerous small trees may decrease the standard deviation even though the stand is diversifying its age structure. A more robust metric of stand complexity might be the standard deviation of basal area classes (grouped into diameter classes), which would be more likely to have a normal distribution than the standard deviation of diameter. There is positive, but somewhat mixed evidence for the benefit of thinning stands as a way to increase stand structural diversity. Given that the pilot study at Redwoods National Park also found benefits to thinning, this result is expected (R. N. Park, 2008). Determining which scenarios lead to the most complex forest structure is important in order to make recommendations for future restoration projects. Based on the results shown in Figure 9, Standard Deviation may not be a reliable indicator of forest structure. The best treatments are seen to be those which keep basal area in a reasonable range and maximize the stem diameter of remaining stems. Predicted basal area is inversely related to treatment intensity (percent of stems removed per year) (Figure 10). Scenarios with thinning mortality greater than 2% per year result in basal areas which are below 150 ft2/ac (see Figure 10), a level likely to harm the stand rather than benefit it, because intact stands should reach high basal areas in maturity. We conclude, therefore, that the ideal thinning scenario should have a total thinning mortality of 2% or less. To simplify consideration of different treatment scenarios on predicted maximum stem diameter, treatment intensity scenarios are normalized to average annualized removal rate. Higher intensity treatments appear to increase the maximum stem diameter. Treatment scenarios with an annualized intensity greater than 2% have the same average diameter, and annualized treatment intensities between 1% and 2% have diameters which are smaller by only 0.5” (See Figure 11). Annualized treatment intensities less than 1% have maximum diameters that are 2” to 6” smaller than the highest intensity treatments (Figure 11). This suggests that annualized treatment intensity should be kept in the 1 – 2% per year range. Light is likely to be the limiting factor for diameter growth in dense stands, however limited nutrients and biophysical constraints may be important at higher light levels. The benefits of increasing light are seen in the initial rise of maximum diameter as thinning intensity rises. The plateau in maximum diameter after thinning intensities greater than 2% per year occurs because light is no longer the limiting factor in diameter growth. Holding the overall intensity of treatment constant at 1% per year, I consider the dependence on maximum diameter on the percentage of trees removed per treatment (Figure 12). Removing more trees per treatment leads to larger trees. The effect of removing more trees per treatment peaks at removals of 40 percent, though removing 30 of 50 percent results in a maximum diameter of only 0.5” to 1” below the peak. Removing a fewer trees per treatment, results in less light availability at the beginning of treatment compared to removing more trees. This means that the remaining trees will grow faster at the beginning of the treatment under the higher intensity treatments even at the same overall annualized thinning intensity. Over the course of 100 years, this difference in initial growth rates results in greater diameter growth in treatments which remove more trees per treatment. Unlike the other metrics analyzed, the standard deviation of stem diameter does not appear to be affected by the intensity of treatment (Figure 13). Given the uncertainty about whether this metric reflects the amount of regrowth or the distribution in stem size, it is unlikely to be useful in determining the best restoration approach. By combining the ideal ranges for thinning intensity and tree removals per treatment, we predict a set of ideal thinning treatments. These treatments have an overall intensity of 1% to 2% average removal per year while removing 30% - 50% of stems at a given treatment. Table 1 shows the 8 treatments which meet these criteria. Notice that all of the selected treatments have long 30 to 50 year periods between treatments. This prediction suggests a middle ground between the current approaches at LEWI (treatment every 20 or 30 years, with removals of 20 – 30% of trees per treatment, (L. a. C. N. H. Park, 2011) and the Ellsworth Creek and Redwoods National Park approach of pre-commercial thinning intensities of 50% to 70% removal (target 100 to 200 trees per acre) once (R. N. Park, 2008). Figure 10: Basal Area after 100 years of treatment versus the annual mortality rate due to thinning. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario. Figure 11: Maximum diameter at age 140 compared with the amount of trees removed per year. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario. Figure 12: Maximum diameter at age 140 for scenarios with an annual removal intensity of 1% as a function of the percentage of trees removed per treatment. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario. Figure 13: Standard Deviation of Diameter at age 140 compared to as a function of the annualized thinning removal. The red line is the stand average from lognormal background scenario and the orange lines are the 95% bootstrapped confidence intervals from the background scenario. Table 1: Suggested treatments (characterized by proportion of trees cut per treatment and frequency of treatment in units of 1/years) for achieving the best stand structure after 100 years indicated in blue highlighted cells of table below. 0.1 Frequency 0.15 Proportion of trees cut 0.2 0.3 0.4 0.5 10 20 30 40 50 Next Steps While the model created for this analysis improved the basic FVS model in several important ways, much more improvement is possible. Analysis could be run on small gridded cells to better predict the effects of local light levels. This could be scaled up and run on an entire landscape. A new version of FVS is being developed in Python, which would allow integration with ArcPY and allow the landscape level simulation to occur. Another improvement would be to let the probability of windthow depend on stand variables in the current stand and neighboring stands. This would allow trees to increase or decrease their windthrow susceptibility, and it would allow catastrophic windthrow to propagate into neighboring cells. Adding these three improvements together would allow one to create and run a landscape FVS simulation which would offer much greater prediction in determining stand composition and structure. One additional improvement would be to take the treatments tested in this research and vary the length over which they were applied to see how long they need to be applied to receive optimal benefits. In addition, it would be interesting to see if an up ramping or down ramping treatment regime would increase structural complexity. Current forests in coastal Oregon and Washington have many other age and species compositions. Therefore, it would be interesting to see whether these treatment options would work for different types of forests. Even on the property owned by LEWI, there are stands of 10 ages with four dominant species compositions. This study only looked at one age and composition of forest. Many of the characteristics of old growth forests are not directly predicted by forest growth models. Therefore, the metrics that I used are proxies for a mature condition. There is no built in function to predict nurse logs, snag dynamics, or CWD dynamics accurately. While FVS does display some estimates for snag number and coarse woody debris, the mathematics of decay is problematic, particularly for large diameter snags and coarse woody debris. Such materials are the most valuable to wildlife and nurse tree perspective and therefore the bias of FVS will likely lead to imperfect results. It would be possible to correct the estimates that FVS gives if more accurate rates of decay were known. Developing this code would be quite difficult and relative to what was already scripted in R, this would make the existing code quite large and unwieldy. A future analysis would ideally include some estimate of these factors. Finally, there is no discussion of cost in this analysis and this means that it becomes harder to make the case for one treatment option versus another. The current approach taken by LEWI uses shorter, more frequent treatments which might cost more than the approach proposed here (depending on the cost structure and discount rates) and it would probably lead to lower complexity both within and between stands. Thinning across a landscape Diversity of structure between stands is another important attribute of natural landscapes whose development is not controlled for timber production. Structural diversity at the landscape scale means that the structure of stands varies across the landscape. This occurs in natural stands because of an overlapping patchwork of natural disturbances. For restoration efforts, more benefit will be gained if the thinning intensity and timing vary across stands. Thinning all of the forest in one year will lead to a lower landscape complexity than varying the thinning intensity across space and time. Landscape heterogeneity is important for maintaining diverse habitats for wildlife habitat and for supporting a diverse understory community (Alaback, 1982). Windthrow While the windthrow occurrence rates used in this analysis are probably reasonable, the exact probabilities of windthrow are not known and they probably vary across the landscape. This is likely a result of several factors: Windthrow depends primarily on the wind shear that trees experience. Wind shear is affected by local wind speed (which is a function of the wind speed above the canopy) tree height, and crown shape. Trees can resist the effects of wind throw based on their height, canopy shape, diameter, and bole strength, rooting, and soil characteristics. Therefore wind throw is actually a function of tree characteristics, stand structure, substrate and wind. Stands can change in their resistance to wind over time, complicating the task of assigning one probability distribution for windthrow. In old growth forests, complete blowdown appears to be less common than the mortality of selected trees within a stand (L. a. C. N. H. Park, 2011). The log-normal distrubution may be the most accurate for the landscape as a whole, but exposed ridges and coastal bluffs are more likely experience severe windthrow more often, and protected valleys are more likely to experience severe windthrow less often. Conclusion Based on a model of forest restoration options, thinning young even aged forests is predicted to be effective at increasing structural complexity across a broad range of treatments. It is suggested that treatments not remove on an annualized average more than 2% of the stand per year in order to maintain an adequate basal area above 150 ft2/ac. Less frequent, higher intensity treatments appear to increase the maximum tree diameter reached in the stand more than lighter, more frequent treatments. Therefore, we recommend that forests use a treatment regime in the following range: thin every 30 years felling 20, 30, or 40% of stems, thin ever 40 years removing 30, 40, or 50 percent of stems, or thin every 50 years and remove 40 or 50 percent of stems. References Alaback, Paul B. (1982). Dynamics of Understory Biomass in Sitka Spruce-Western Hemlock Forests of Southeast Alaska. Ecology, 63(6), 1932-1948. doi: 10.2307/1940131 Baldocchi, Dennis D. (2003). Assessing the eddy covariance technique for evaluating carbon dioxide exchange rates of ecosystems: past, present and future. Global Change Biology, 9(4), 479492. Boit, John, Howay, F. W., Elliott, T. C., & Young, F. G. (1921). John Boit's Log of the Columbia—17901793. The Quarterly of the Oregon Historical Society, 22(4), 257-351. doi: 10.2307/20610191 Boone, Richard D., Sollins, Phillip, & Cromack, Kermit, Jr. (1988). Stand and Soil Changes Along A Mountain Hemlock Death and Regrowth Sequence. Ecology, 69(3), pp. 714-722. Cole, D. W., Gessell, S. P., & Dice, S. F. (1968). Paper presented at the Proceedings of the 13th Annual Meeting for the American Association for the Advancement of Sciences. Crookston, Nicholas L., & Dixon, Gary E. (2005). The Forest Vegetation Simulator: A review of its structure, content, and applications. Computers and Electronics in Agriculture, 49(1), 60-80. Forsman, Eric D., Meslow, E. Charles, & Wight, Howard M. (1984). Distribution and Biology of the Spotted Owl in Oregon. Wildlife Monographs(87), 3-64. doi: 10.2307/3830695 Graham, Robin Lee Lambert. (1981). Biomas dynamics of dead douglas-fir and western hemlock boles in mid-elevation forests of the Cascade Range. Oregon State University. Grier, Charles C., & Logan, Robert S. (1977). Old-Growth Pseudotsuga menziesii Communities of a Western Oregon Watershed: Biomass Distribution and Production Budgets. Ecological Monographs, 47(4), pp. 373-400. Hamer, Thomas E., & Nelson, S. Kim. (1995). Characteristics of marbled murrelet nest trees and nesting stands. Ecology and Conservation of the Marbled Murrelet. USDA Forest Service General Technical Report PSW-GTR-152. Pacific Southwest Research Station, Albany, CA, 6982. Harmon, M. E., Franklin, J. F., Swanson, F. J., Sollins, P., Gregory, S. V., Lattin, J. V., . . . Cummins, K. W. (1986). Ecology of Coarse Woody Debris in Temperate Ecosystems. Paper presented at the Recent advancements in ecological research. Harmon, Mark E., Ferrell, William K., & Franklin, Jerry F. (1990). Effects on Carbon Storage of Conversion of Old-Growth Forests to Young Forests. Science, 247(4943), pp. 699-702. Harmon, Mark E., & Hua, Chen. (1991). Coarse Woody Debris Dynamics in Two Old-Growth Ecosystems. BioScience, 41(9), 604-610. Hudiburg, Tara, Law, Beverly, Turner, David P., Campbell, John, Donato, Dan, & Duane, Maureen. (2009). Carbon Dynamics of Oregon and Northern California Forests and Potential LandBased Carbon Storage. Ecological Applications, 19(1), pp. 163-180. Kurz, Werner A., & Apps, Michael J. (1999). A 70-Year Retrospective Analysis of Carbon Fluxes in the Canadian Forest Sector. Ecological Applications, 9(2), pp. 526-547. Landsberg, J. J., Waring, R. H., & Coops, N. C. (2003). Performance of the forest productivity model 3PG applied to a wide range of forest types. Forest Ecology and Management, 172(2–3), 199214. doi: 10.1016/S0378-1127(01)00804-0 Lewis, M., Clark, W., and Members of the Corps of Discovery. (2002). The Journals of the Lewis and Clark Expedition (G. Moulton, Ed.). Retrieved 9/30, 2013, from http://lewisandclarkjournals.unl.edu/ Liane Davis, Yoav Bar-Ness, Travis Woolley, David Shaw, and David Rolph. (2009). Characterization of biological diversity, structure, and composition within old-growth forest refugia and young-managed forests in the Willapa Hills, Washington: The Nature Conservancy. Luyssaert, Sebastiaan, Schulze, E. Detlef, Borner, Annett, Knohl, Alexander, Hessenmoller, Dominik, Law, Beverly E., . . . Grace, John. (2008). Old-growth forests as global carbon sinks. Nature, 455(7210), 213-215. Nowacki, Gregory Jay, & Kramer, Marc G. (1998). The effects of wind disturbance on temperate rain forest structure and dynamics of southeast Alaska: US Department of Agriculture, Forest Service, Pacific Northwest Research Station. Odum, Eugene P. (1969). The Strategy of Ecosystem Development. Science, 164(3877), pp. 262-270. Park, Lewis and Clark National Historic. (2011). Fort Clatsop Unit Forest Restoration Plan: Environmental Assessment Park, Redwoods National. (2008). South Fork Lost Man Creek Second Growth Forest Restoration Environmental Assessment. Schreiner, Edward G., Krueger, Kirsten A., Houston, Douglas B., & Happe, Patricia J. (1996). Understory patch dynamics and ungulate herbivory in old-growth forests of Olympic National Park, Washington. Canadian Journal of Forest Research, 26(2), 255-265. Smithwick, Erica A. H., Harmon, Mark E., Remillard, Suzanne M., Acker, Steven A., & Franklin, Jerry F. (2002). Potential Upper Bounds of Carbon Stores in Forests of the Pacific Northwest. Ecological Applications, 12(5), pp. 1303-1317. Taylor, Alan H. (1990). Disturbance and Persistence of Sitka Spruce (Picea sitchensis (Bong) Carr.) in Coastal Forests of the Pacific Northwest, North America. Journal of Biogeography, 17(1), 47-58. doi: 10.2307/2845187 Turner, J., & Long, James N. (1975). Accumulation of Organic Matter in a Series of Douglas-fir Stands. Canadian Journal of Forest Research, 5(4), 681-690. Appendix A: Historical Analysis Methods Historical Analysis The ecological history of the forests of the Columbia has not been written in one definitive guide. Therefore, research in primary sources is necessary to better understand the landscape which Lewis and Clark saw around Fort Clatsop and the lower Columbia River. I went through several primary sources in order to piece together a picture of the landscape of 1805-1806. In addition, I looked at these sources to see if that landscape was significantly changed by the actions of native people and European traders who frequented the mouth of the Columbia. Referring back to historical sources is important in this case because the Environmental Assessment that Lewis and Clark National Historical Park produced explicitly tied the restoration goal to the conditions witnessed by Lewis and Clark. Yet that document did not show a thorough analysis to determine what this baseline would have looked like. Instead, they assumed that it would be old growth forests free from anthropogenic disturbance. This is probably a reasonable assumption, but it ignores the more subtle and potentially important effects that native people and traders may have had. I read Lewis and Clark’s Journal from their stay at fort Clatsop and their travels on the lower Columbia River to determine what direct and indirect effects native people might have had on the natural environment. I interpreted the historical descriptions (written by Lewis and Clark) of the landscapes along the lower Columbia River to produce a reconstruction of the historical landscape of 1805-1806. This map was produced from a combination of maps that the expedition created and interpretation of current landforms and their probable historical cover type. This reconstruction is reasonably accurate, but the details are definitely imperfect in their accuracy. This arises from two separate factors: Lewis and Clark did not specifically map the environment, they did incorporate natural features onto existing maps, but this coverage was not systematic and therefore much interpretation is required. In addition, there has been much alteration of the natural hydrology from a combination of recent land accretion and diking around farmland. Therefore, the exact extents of these historical land uses are probably impossible to determine or verify. But in such a dynamic area, there was probably much change from year to year, and an educated guess certainly improves our understanding even with known caveats about accuracy. Historical Results Historical Maps Lewis and Clark certainly created many maps, particularly Clark, who acted as the expedition’s cartographer. These maps are of immense value; however they were made with the express purpose of surveying a new landscape for navigation. The details of natural ecosystems are not consistently included. Yet Lewis and Clark recorded copious descriptions of the locations that they visited, and therefore it is possible to learn more about the landscapes encountered by the expedition through these written descriptions. I set out to use these narrative descriptions as the basis of a historical map of the landscape directly adjacent to Fort Clatsop and the land near the present day Lewis and Clark National Historical Park. Figure 15 shows the descriptions that Lewis and Clark used to describe specific features of the landscape along the lower Columbia River. The general trend is that the Columbia and Netul rivers had significant wetlands along their shores. Dense old growth forests dominated the rugged coastal hills, and rolling coastal prairies stretched along the entire sea shore, but were most extensive from Point Adams down south along the shoreline west of Fort Clatsop. These general patterns were drawn on a map to represent a possible landscape configuration in 1805-1806 (Figure 14). One significant change in addition to the land accretion is that much wetland area has been lost due to levees, dikes, and changes to the hydrology of the coastal landscape from Cape Adams to Tillamook Head. Specific Land Cover Types Forests Lewis and Clark describe forests with significant old growth characteristics covering most of the hills and rugged terrain near the mouth of the Columbia River. They describe forests composed of Sitka spruce (Picea sitchensis), western hemlock (Tsuga heterophylla), douglas fir (Pseudotsuga menziesii), red alder (Alnus rubra), red cedar (Thuja plicata), vine maple (Acer circinatum), and bigleaf maple (Acer macrophyllum). Clark describes the general character of coastal forests in the following passage. The vegetable productions are also numerous. The hills along the coast are high and steep, and the general covering is a growth of lofty pines of different species, some of which rise more than two hundred feet, and are ten or twelve feet in diameter at the root. Besides these trees we observe on a point a species of ash, the alder, the laurel, one species of the wild crab, and several kinds of underbrush, among which rosebushes are conspicuous. Journals of Lewis and Clark, November 29, 1805 (Lewis, 2002) The main forest trees, Sitka spruce, Western hemlock, and douglas fir reached impressive sizes in most forests on the coastal hills. Lewis describes the trees south of Tillamook head in the following passage. The country in general as about Fort Clatsop is covered with a very heavy growth of several species of pine & furr, also the arbor vita or white cedar and a small proportion of the black Alder which last sometimes grows to the hight of sixty or seventy feet, and from two to four feet in diameter. some species of the pine rise to the immence hight of 210 feet and are from 7 to 12 feet in diameter, and are perfectly sound and solid. Journals of Lewis and Clark, January 8th, 1806 (Lewis, 2002) Lewis and Clark also noted several features of old growth forests such as nurse logs: In many places pine trees, three or four feet in thickness are seen growing on the bodies of large trees, which though fallen and covered with moss were in part sound. Journals of Lewis and Clark, November 17th, 1805 (Lewis, 2002) Soils are described as having thick organic and A horizons: “though the land in general possesses a thick black mould” [Journals of Lewis and Clark, December 28th, 1805 (Lewis, 2002)]. Certain areas had younger forests which were not in old growth condition. In the following passage, Lewis describes the vegetation in the lowlands west of Cape Disappointment. The whole lower country was covered with almost impenetrable thickets of small pine, with which is mixed a species of plant resembling arrowwood, twelve or fifteen feet high, with a thorny stem, almost interwoven with each other, and scattered among the fern and fallen timber…This thick growth rendered travelling almost impossible. Journals of Lewis and Clark, November 13th, 1805 (Lewis, 2002) Sitka Spruce is described as largest tree species, both in height and diameter of the common trees of the coastal land near the mouth of the Columbia River. The first species grows to an immense size, and is very commonly twenty seven feet in circumference six feet above the earth’s surface: they rise to the height of two hundred and thirty feet, and one hundred and twenty of that height without a limb. We often found them thirty-six feet in circumference. One party measured one, and found it to be forty-two feet in circumference, at a point beyond the reach of an ordinary man. This trunk for the distance to two hundred feet was destitute of limbs: this tree was perfectly sound, and at a moderate calculation, its size may be estimated at three hundred feet. Journals of Lewis and Clark, Chapter 21, page 383 (Lewis, 2002) Lewis also describes western hemlock and importantly, he gives some information about its abundance. The second is a much more common species, and constitutes at least half of the timber in this neighborhood. it seems to resemble the spruce, rising from one hundred and sixty to one hundred and eighty feet, and it is from four to six in diameter, straight, round, and regularly tapering. Journals of Lewis and Clark, Chapter 21, page 383 (Lewis, 2002) Douglas fir is described as “a species of fir which arrives to the size of Nos. 2 and 4, the stem simple branching, diffuse and proliferous” [February 6th, 1806, (Lewis, 2002)]. The understory species of maple, bigleaf and vine, are described as: This tree [Bigleaf maple] is frequently 3 feet in diameter and rises to 40 or 50 feet high. The fruit is a winged seed somewhate like the maple. in the same part of the country there is also another growth [Vine Maple] which resembles the white maple in it's appearance, only that it is by no means so large; seldom being more than from 6 to 9 inches in diamater, and from 15 to 20 feet high; they frequently grow in clusters as if from the same bed of roots spreading and leaning outwards. ” Journals of Lewis and Clark, February 6th, 1806, (Lewis, 2002) From these passages, we can begin to reconstruct the nature of the forests as Lewis and Clark observed in 1805. Clearly, much of the forest was still in an old growth condition with large trees up to 13.3’ in DBH. Sitka spruce remained the king of coastal forests, reaching the largest diameter and the highest height. Western hemlock comprised greater than 50% of tree cover and was the most common tree species. Douglas fir was found both in the forested hills, and on small rises in wetlands. The hardwood species, red alder, vine maple, and bigleaf maple were locally important components of forests. Red cedar is conspicuously missing, except for the forests of Ecola State park, on the southern end of Tillamook Head. Cedar is usually considered a common species of coastal forests by most modern range maps. Yet it appeared to be absent form many of the forests along the lower Columbia. There is no good ecological reason why it would not be found in these forests. Interestingly, the first descriptions of the forest by John Boit are somewhat contrary to what Lewis and Clark Described. He says that “the woods [are]mostly clear from Underbrush… We found plenty of Oak, Ash, and Walnut trees” [Journal of John Boit, May 18th, 1792 (Boit, Howay, Elliott, & Young, 1921)]. None of Lewis and Clark’s descriptions make mention of woods clear of underbrush. This observation may just represent one place where Boit and Gray Landed, and therefore we will take the descriptions of Lewis and Clark as being more accurate. Coastal Prairies The land directly adjacent to coastal dunes and cliffs is described as being covered by a coastal prairie. Such a feature is described as being “formed of open waving prairies of sand, with ridges running parallel to the river, and covered with green grass” [Journals of Lewis and Clark, December 28th, 1805 (Lewis, 2002)]. On areas with coastal cliffs such as Cape Disappointment and Tillamook head, these features were “covered with thick timber on the inner side, but open and grassy in the exposure to the sea” [Journals of Lewis and Clark, November 17th, 1805 (Lewis, 2002)]. The expedition left little record of the species composition on these prairies, probably because they were relatively unremarkable to the expedition. John Boit, a crew member on Columbia led by the merchant Robert Gray described a potential prairie like landscape somewhere between station camp and Cape Disappointment in the following manner: “Found much clear ground, fit for Cultivation” [Journal of John Boit, May 15th, 1792 (Boit et al., 1921)]. Wetlands and riparian corridors Wetlands and riparian habitats were much more plentiful in 1805 than they are today. These areas were split into three types of wetlands: freshwater wetlands and bogs, riparian areas, and brackish salt marshes. The freshwater bogs were described by members of the expedition who were hunting elk near the coast, directly west of Fort Clatsop. The ground for a whole acre would shake at our tread, and sometimes we sunk to our hips without finding bottom. Over the surface of these bogs is a species of moss, among which are great numbers of cranberries, and occasionally there rise from the swamp steep and small knobs of earth, thickly covered with pine and laurel. Journals of Lewis and Clark, December 8th, 1805 (Lewis, 2002) The salt marshes to the east of Cape Disappointment were described by Clark, who explored them when the expedition was still at station camp. Clark describes the area as follows “The country is low, open and marshy; interspersed with some high pine and a thick undergrowth” [Journals of Lewis and Clark, November 17th, 1805 (Lewis, 2002)]. Lewis and Clark do not directly describe the riparian areas so we are unsure about what these areas would have been like. However Lewis does refer to the existence of widely spread wetlands in the area of Fort Clatsop as the “swamp of the Netul” [Journals of Lewis and Clark, January 5th, 1806 (Lewis, 2002)] Weather The expedition describes the weather many times during their stay at fort Clatsop due to their intense dislike of the constantly moist coastal climate. On November 15th, 1805 Lewis says that “The rain, which has continued for the last ten days without an interval of more than two hours, has completely wet all of our merchandise”(Lewis, 2002). They also described frequent storms which were so violent as to hinder their travel down the Columbia River. It rained the whole night, and about daylight a tremendous gale of wind rose from the S.S.E. and continued during the whole day with great violence. The sea runs so high that the water comes into our camp, which the rain prevents from leaving. Journals of Lewis and Clark, November 22, 1805 (Lewis, 2002) These storms also caused windthrow in the coastal forests, which Lewis and Clark observed. At noon the wind shifted to the northwest, and blew with such tremendous fury that many trees were blown down around us. This gale lasted with short intervals during the whole night; but towards morning the wind lulled, though the rain continued, and the waves were still high. Journals of Lewis and Clark, November 27, 1805 (Lewis, 2002) Nothing on the expedition had prepared them for the harsh weather and conditions of winters in the coastal range. While they do not directly admit this in the journals, it is clear from their descriptions that Lewis and Clark were not very comfortable dealing with the storms and large waves which are common on the lower Columbia River. It rained and hailed during the day, and high wind from the southeast not only threw down trees as we passed, but made the river so rough that we proceeded with great risk. Journals of Lewis and Clark, February 15th, 1806 (Lewis, 2002) Native uses of natural resources Native people of the Chinook and Clatsop tribes made thorough use of natural resources found near the Columbia River. They caught fish and whales through a variety of means, including spears, nets, and waiting for winter storms to drive fish up on the beach. The Clatsops Chinnooks &c. in fishing employ the common streight net, the scooping or diping net with a long handle, the gig, and the hook and line. the common net is of different lengths and debths usually employed in taking the sammon, Carr and trout in the inlets among the marshey grounds and the mouths of deep creeks. the skiming or [s]cooping net to take small fish in the spring and summer season; the gig and hook are employed indiscriminately at all seasons in taking such fish as they can procure by their means. Journals of Lewis and Clark, January 16th, 1806 (Lewis, 2002) Fur bearing animals were trapped, even before Europeans began trading as described by John Boit. The Indians are very numerous, and appear’d very civill (not even offering to steal). during our short stay we collected 150 Otter, 300 Beaver, and twice the Number of other land furs. the river abounds with excellent Salmon, and most other River fish, and the Woods with plenty of Moose and Deer, the skins of which was brought us in great plenty, and the Banks produces a ground Nut, which is an excellent substitute for either bread or Potatoes. Journals of John Boit, May 18th, 1792 (Boit et al., 1921) Native tribes who lived near the mouth of the Columbia traded extensively because certain foodstuffs were produced in specific locations by tribes who specialized in producing one or two products. There was a type of dry pounded salmon which was made only by native peoples who trapped salmon at the Dalles, a natural falls in the cascade range, in Washington state today. we gave six small fish-hooks, a worn-out file, and some pounded fish which had become soft and mouldy by exposure, that we could not use it: It is, however, highly prized by the indians of this neighborhood. Although a very portable and convenient food, the mode of curing seems known, or at least practised only by the indians near the great falls, and coming from such a distance, has additional value in the eyes of these people, who are anxious to possess some food less precarious than their ordinary subsistence. Journals of Lewis and Clark, December 23rd, 1805 (Lewis, 2002) Wapato roots (the groundnut referred to by Boit) were naturally found in freshwater wetlands. There is a large complex of wetlands 30-50 miles southeast of the mouth of the Columbia River which served as a source for wappato which was traded throughout the region. We purchased from the old squaw for armbands and rings, a few wappatoo roots, on which we subsisted. They are nearly the equal in flavour to the irish potato, and they afford a very good substitute for bread. Journals of Lewis and Clark, November 22nd, 1805 (Lewis, 2002) Members of the Clatsop and Chinook nations made extensive use of western red cedar (Thuja plicata) for canoes, clothing, and shelter. Cedar bark was massaged until it separated into fibers and used to make many articles including baskets which Lewis says “are formed of cedar bark and beargrass so closely interwoven with the fingers that they are watertight without the aid of gum or rosin” [Journals of Lewis and Clark, January 17th, 1806 (Lewis, 2002)]. Cedar was the preferred wood for canoe making, and Lewis observed that the canoes made by the Clatsop and Chinook people were far superior to their own. The natives inhabiting the lower portion of the Columbia River make their canoes remarkably neat light and well addapted for riding high waves. I have seen the natives near the coast riding waves in these canoes with safety and apparently without concern where I should have thought it impossible for any vessel of the same size to lived a minute. they are built of whitecedar or Arborvita generally, but sometimes of the firr. they are cut out of a solid stick of timber, the gunwals at the upper edge foald over outwards and are about ⅝ of an inch thick and 4 or five broad, and stand horrizontally forming a kind of rim to the canoe to prevent the water beating into it … some of the large canoes are upwards of 50 feet long and will carry from 8 to 10 thousand lbs. or from 20 to thirty persons. Journals of Lewis and Clark, February 1st, 1806 [Lewis] (Lewis, 2002) Short bows were made by the Chinookan people out of Cedar and backed with elk sinew, and described by Lewis in the following passage: Their bows are extreamly neat and very elastic, they are about two and a half feet in length, and two inches in width in the center, thence tapering graduly to the extremities where they are half an inch wide they are very flat and thin, formed of the heart of the arbor vita or white cedar, the back of the bow being thickly covered with sinews of the Elk laid on with a gleue which they make from the sturgeon; the string is made of sinues of the Elk also. Journals of Lewis and Clark, January 15th, 1806 (Lewis, 2002) Native trading The Chinook and Clatsop peoples depended heavily on trading both because they desired the productions of neighboring tribes and they were the primary contact native peoples to interface with European fur traders. Clark determined that 13 boats entered to each year to trade [Journals of Lewis and Clark, January 1st, 1806 (Lewis, 2002)], primarily for fur. There is no mention of timber harvesting by these early European and American visitors. The coastal tribes traded fur directly for trade goods, primarily blue beads, which were highly prized. Towards the evening several Clatsops came over in a canoe with two skins of the sea-otter. To this article they attach an extravagant value and their demands for it were so high that we were fearful of reducing our small stock of merchandise, on which we must depend for subsistence as we return, to venture on purchasing. To ascertain however their ideas as to the value of different objects, we offered for one of the skins a watch, a handkerchief, an american dollar, and a bunch of red beads’ but neight the curious mechanism of the watch nor even thered beads could tempt him. he refused the offer but asked for tiacomoshack or chief beads, the most common sort of coarse blue-colored beads, the article beyond all price in their estimation. Journals of Lewis and Clark, November 23rd, 1805 (Lewis, 2002) These blue beads served as a sort of currency between native tribes, and were used to procure food from other tribes. One of the indians had a robe made of two sea-otter skins, the fur of which was the most beautiful we had ever seen; the owner resisted every temptation to part with it, but a length could not resist the offer of a belt of blue beads which Chaboneau’s wife wore around her waist. Journals of Lewis and Clark, November 20th, 1805 (Lewis, 2002) Lessons from the past Lewis and Clark describe forests near Fort Clatsop as thick, with large diameter conifers. They record a forest which was half western hemlock with significant amounts of Sitka Spruce, Douglas Fir, Red Alder, and minor amounts of Vine Maple and Bigleaf Maple. The overstory trees were large, with sitka spruce normally at 6-10’ DBH with some trees at 13 or 14’ in DBH. Western Hemlock was on average 4-6’ in diameter, and the other species forest trees reached similarly impressive sizes. One species which was conspicuously absent was western red cedar. Cedar was only found by the expedition on the south side of Tillamook head. The reason for this is not totally clear, but the best explanation probably involves heavy use and depletion by native peoples of the lower Columbia River. The strong importance of cedar to native peoples and the difficulty of transporting logs over great distances without modern technology would probably have put high pressure on red cedar near water. After 100s to 1000s of years of native use, it is likely that Western Red Cedar was less common in coastal forests near the lower Columbia than it would have been without the anthropogenic influence. Animal and fish populations were high though slightly depleted during the time that Lewis and Clark stayed at Fort Clatsop. John Boit describes buying roughly 900 from native peoples who had never traded with westerners, and hence were not involved in the fur trade. Lewis and Clark happened on a landscape which had seen 10 years of heavy trading in fur and therefore it is likely that fur bearing animals were much less plentiful compared to when Robert Gray’s expedition entered the Columbia. The change in elk density may have affected forest regrowth and succession due to changes in browsing pressure. Current studies have shown that elk can slow succession and perpetuate grassy clearings in Olympic National Park (Schreiner, Krueger, Houston, & Happe, 1996). Much of the landscape along the Columbia River in 1805 was covered by wetlands and salt marshes. The Columbia Estuary was had extensive wetlands along the southern shore with the exception of rocky outcropping which underlies Astoria. Many of the small valleys which drained the coastal hills near fort Clatsop had riparian wetlands in their valleys. On the western side, there were extensive wetlands which Lewis and Clark describe with dislike. These wetlands would have provided important habitat for anadromous fish such as Salmon and Sturgeon. Yet the landscape of today is much changed from its historical past. While forests still cover much of the landscape, they are primarily forests of western hemlock and Douglas fir. Sitka Spruce is less common today than it would have been in 1805. Nearly all of the old growth forests have been logged, and finding large Sitka Spruce is uncommon. Most of the wetlands have been diked and filled, or affected by sedimentation along the jetties which protect the mouth of the Columbia River. Elk and fur bearers still exist in this matrix of young forests, agriculture and development, but trapping and habitat loss have likely decreased their numbers from pre-European times. Figure 14: Reconstructed Landuse directly adjacent to Fort Clatsop. Notice the large amount of land accretion beyond Point Adams. This reconfiguration of the mouth of the Columbia has significantly protected the lower river from ocean swells. Figure 15: Lewis and Clark’s Description of the landscape of the Lower Columbia River in 1805-1806. Locations are approximate, and in some cases, may refer to whole landscape features beyond the point on the map. Appendix B: Carbon Sequestration Carbon Literature Review The data problem Classical views on forest growth hold that forests accumulate carbon only in their early exponential growth phases. This view was promulgated to Odum (Odum, 1969) and yet has continued to be the standard until recently. Part of this was caused by the available forest growth data, which was mainly in young managed stands. Money was not available for repeat sampling of old growth stands, and in those stands, the above ground biomass was well measured, but soil and root carbon was not well known. Some of the best data available comes from the Forest Inventory Analysis (FIA) program run by the US Forest service. Most of these plots are in young managed timber stands, though some are in older plots. In general, data focuses on young forests, which do show a slowing of growth in maturity when the fast growing conifer trees begin to experience density dependent mortality. This leads to a model of forest growth based primarily on even aged conifer trees grown in clear cuts. Old growth forests are much more complex in their development occurs over much longer time frames because individual trees live for several human lifetimes. Recent changes in our understanding of carbon stored in old growth forest have come from the eddy flux network. Eddy flux towers measure gas exchange and can therefore directly measure carbon uptake in a way which direct measurements of living, dead and soil pools cannot do (Baldocchi, 2003). . Additionally, there is the problem that forests do not function as steady state systems. Carbon storage and growth change dramatically from establishment to old growth. Even in mature forests, periodic disturbances, which may occur at return intervals of 100s to 1000s of years, play important roles in altering the current and future balance of carbon in a stand. Therefore any model which presents one rate inherently misses the fact that forests are dynamic and ever changing and we may get the rates wrong because we are only looking at temporary fluxes. The classical understanding of forest carbon dynamics Odum studied young forest ecosystems and used a test tube community of bacteria as an analogue for the growth and succession of a forest ecosystem. He postulated that ecosystems grew exponentially for the first 30 years and then experienced declining growth until a biomass maximum which was reached at 80 years old (Odum, 1969). Odums ideas about biomass accumulation assume a rapid turnover of dead organic matter and a system which quickly reaches equilibrium. Even aged stands grown for timber do seem to show this trend. Many of the forest growth predictions and policies about carbon storage implicitly assume that maximum biomass will be reached by mature and old growth forests. Recent revisions to old growth forests role in carbon storage Luyssaert conducted a thorough review of old growth forests and concluded that old forests were definite carbon sinks (Luyssaert et al., 2008). More importantly, stem growth accounted for the least important part of this carbon sequestration at 0.4 MgC/ha. Coarse woody debris and soil accounted for much more of the carbon sequestered at 0.7 MgC/ha and 1.4 MgC/ha (Luyssaert et al., 2008). His analysis updated the ideas set forth by Odum and therefore changed our understanding of forest ecology and carbon cycling. This work also highlighted the importance of retaining and restoring old growth conditions both for the ecological and climate mitigation value that they provide. One of the ideas promoted by Luyssart and other authors is that when larger trees die, they decompose more slowly than regrowth occurs, potentially maintaining carbon storage for 10s to 100s of years (Boone, Sollins, & Cromack, 1988; Graham, 1981; M. E. Harmon et al.; Mark E. Harmon & Hua, 1991; Luyssaert et al., 2008). Luyssart calculated that old growth forests sequester 2.4 MgC Ha-1 yr-1, which when scaled up to the 6 x 108 hectares of primary forest, yields a sink of 1.5 0.3 GtC yr-1 (Luyssaert et al., 2008). This result strongly suggests that old growth fores ts are critical to carbon storage and sequestration. Therefore, forest restoration has the potential to play a role in the carbon storage if restored forests act like old growth forests. Old Growth Forests, Young Forests and Forest restoration In order to understand the bounds on forest restoration, I conducted a literature review of carbon storage in both young and old growth forests in the Pacific Northwest (table 1, table 2). Young forests have received more study, but recent work has illuminated the role that old forests have in storing forest carbon. In terms of carbon content, old and young stands differ markedly in their allocation of carbon and the size of the total pool. The greatest differences in carbon storage are seen in stem mass where old growth forests store 323 MgC/ha (Grier & Logan, 1977)and young forests store 145 MgC/ha (Turner & Long, 1975). Old growth branches store 27.3 MgC/ha (Grier & Logan, 1977) while young stands store 7 MgC/ha (Turner & Long, 1975) in their branches. Foliage and fine roots store similar amounts of carbon in both young and old stands (Grier & Logan, 1977; Mark E. Harmon, Ferrell, & Franklin, 1990; Turner & Long, 1975). Coarse woody debris in old growth stands can contain 10 times as much carbon as young stands (Cole, Gessell, & Dice; Mark E. Harmon et al., 1990). Estimates from across Canada show that storage across physiographic province differs substantially (Kurz & Apps, 1999). The temperate coastal ecosystems from central California to southern Alaska store much more carbon than other ecosystems. These landscapes may store a maximum of 700 to 1200 MgC/ha, which is one higher than most other forest types. (Luyssaert et al., 2008; Smithwick, Harmon, Remillard, Acker, & Franklin, 2002). Hudiburg conducted an analysis of forests of three ages: young (<80 years old), mature (80 - 200 years old), and old (>200 years old), and found that carbon storage was highest in the oldest stands (Hudiburg et al., 2009). From the range of work which has already been done on forest carbon storage, we can see that old growth forests store roughly 2 to 3 times as much carbon as young forests in the pacific northwest. The coastal Sitka Spruce and Redwood forests are some of the most productive both in terms of forest growth and carbon storage. Compared to other forests, these pacific coastal forests have a greater differential between maximum carbon storage of old forests compared to the storage in younger forests (Smithwick et al., 2002). Therefore, forest undergoing restoration should store more carbon as the forest matures, particularly in the Pacific Northwest. Table 2: Thinning scenarios evaluated to determine the resulting amount of carbon stored in tons/acre. The columns correspond to the proportion of trees cut at every thinning, and the rows denote the return interval of the thinning. Tons/acre 0 868 10 years 20 years Frequency 30 years 40 years 50 years 0.1 668 750 807 806 811 0.15 610 694 780 780 785 Proportion of trees cut 0.2 0.25 0.3 578 563 560 649 614 591 749 723 696 750 719 697 756 731 705 0.4 566 566 649 649 658 0.5 577 564 611 611 620 Table 3: Comparison of published rates of change for carbon stored in old growth forests Pool Tree Stems CWD Soil + Roots Total Rate of change 0.4(0.1) MgC/ha/yr 0.7(0.2) MgC/ha/yr 1.3(0.8) MgC/ha/yr 2.4 (0.8) MgC/ha/yr 1.7(0.2) MgC/ha/yr 1.7-2.9 MgC/ha/yr 1.7 (1.0) Reference Lyussart Lyussart Lyussart Lyussart Grier and Logan 1977 Notes Summary of many other rates Summary of many other rates Summary of many other rates Summary of many other rates Wild River Old Growth (500 years old) Forest, WA (eddyFlux based) Medium aged forests (80-200 years old) (eddy flux based) Table 4: Comparison of published estimates of carbon pools in old growth forests Global Old growth Carbon Pools Category Foliage Branches Soil CWD FWD Coarse roots Fine roots Stems Living Biomass Dead biomass Total 3 = Edmonds and Mara 2 = Smithwick 2002 16 = Grier and Logan 1977 40 = Harmon1990 25 = Harmon1986 1 = Kurz and Apps 1999 4 = Hudiburg 2009 7 = Lyaessart 2008 Notes (location) Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Olympic Natl. Park, Hoh valley Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Old (>200 years old), Oregon and California Coast Range Old (>200 years old), Western Cascades Old (>200 years old), Sierra Nevada Old (>200 years old), Klamath Mountains Mature (80 - 200 years old), Oregon and California Coast Range Mature (80 - 200 years old), Western Cascades Mature (80 - 200 years old), Sierra Nevada Mature (80 - 200 years old), Klamath Mountains Old (>200 years old), Oregon and California Coast Range Old (>200 years old), Western Cascades Old (>200 years old), Sierra Nevada Old (>200 years old), Klamath Mountains Mature (80 - 200 years old), Oregon and California Coast Range Mature (80 - 200 years old), Western Cascades Mature (80 - 200 years old), Sierra Nevada Mature (80 - 200 years old), Klamath Mountains Doug Fir and Hemlock Forest (450 years old) Old (>200 years old), Oregon and California Coast Range Old (>200 years old), Western Cascades Old (>200 years old), Sierra Nevada Old (>200 years old), Klamath Mountains Mature (80 - 200 years old), Oregon and California Coast Range Mature (80 - 200 years old), Western Cascades Mature (80 - 200 years old), Sierra Nevada Mature (80 - 200 years old), Klamath Mountains Maximum storage in General Old growth maximum storage in Coastal Oregon Pools (reference) 6.2 - 7.0 MgC/ha 26.3 MgC/ha 56 MgC/ha 97 MgC/ha 205 MgC/ha 26 MgC/ha 71 MgC/ha 5.6 MgC/ha 323 MgC/ha 300(145) MgC/ha 221(119) MgC/ha 149(96) MgC/ha 200(119) MgC/ha 227(124) MgC/ha 146(98) MgC/ha 111(73) MgC/ha 130(89) MgC/ha 47(48) MgC/ha 49(41) MgC/ha 27(22) MgC/ha 26(23) MgC/ha 36(29) MgC/ha 25(22) MgC/ha 19(17) MgC/ha 16(15) MgC/ha 611 - 612 MgC/ha 347(104) MgC/ha 270(160) MgC/ha 176(118) MgC/ha 226(142) MgC/ha 263(153) MgC/ha 171(120) MgC/ha 130(90) MgC/ha 146(104) MgC/ha 700 MgC/ha 1127 MgC/ha Reference Grier and Logan 1977, Harmon et al 1990 Grier and Logan 1977 Grier and Logan 1977 Harmon et al 1986 Harmon et al 1986 Grier and Logan 1977 Grier and Logan 1977 Grier and Logan 1977 Grier and Logan 1977 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Harmon 1990, Grier and Logan 1977 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Luyssaert 2008 Smithwick 2002 Table 5 Comparison of published estimates of carbon pools young forests Young Even Aged Stands Category Foliage Branches Soil CWD FWD Coarse roots Fine roots Stems Aboveground Biomass Living Biomass Belowground Biomass Dead Organic Matter Total Notes (location) Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Doug Fir and Hemlock Forest Pools (Standard Deviation) 5.5 MgC/ha 7 MgC/ha 56 MgC/ha 3.8 MgC/ha 19 MgC/ha Doug Fir and Hemlock Forest 7.1 MgC/ha Doug Fir and Hemlock Forest 29 MgC/ha Doug Fir and Hemlock Forest 5.6 MgC/ha (35) Doug Fir and Hemlock Forest 145 MgC/ha (20) Canada - general - Medium aged 28.4 MgC/ha (1) Canadian Boreal west 25.2 MgC/ha (1) Canadian Boreal east 15.3 MgC/ha (1) Canadian sub-arctic 14.1 MgC/ha (1) Canadian Cool temperate 44.1 MgC/ha (1) Canadian Moderate temperate 124.9 MgC/ha (1) Canadian Cordilleran 56 MgC/ha (1) Canadian Interior cordilleran 70.5 MgC/ha (1) Pacific Cordillera 115.5 MgC/ha (1) Sub-arctic cordillera 10 MgC/ha (1) Young(<80 years old), Oregon and California Coast Range 109 (69) MgC/ha(4) Young(<80 years old), Western Cascades 62 (52) MgC/ha(4) Young(<80 years old), Sierra Nevada 38 (35) MgC/ha(4) Young(<80 years old), Klamath Mountains 52 (50) MgC/ha(4) Canada - general 7.5 MgC/ha (1) Canadian Boreal west 7.2 MgC/ha (1) Canadian Boreal east 4.4 MgC/ha (1) Canadian sub-arctic 3.9 MgC/ha (1) Canadian Cool temperate 10.5 MgC/ha (1) Canadian Moderate temperate 25 MgC/ha (1) Canadian Cordilleran 14.1 MgC/ha (1) Canadian Interior cordilleran 17.1 MgC/ha (1) Canadian Pacific Cordillera 27.3 MgC/ha (1) Canadian Sub-arctic cordillera 3.3 MgC/ha (1) Canada - general - Medium aged 176.5 MgC/ha (1) Canadian Boreal west 138.3 MgC/ha (1) Canadian Boreal east 145.1 MgC/ha (1) Canadian sub-arctic 248.1 MgC/ha (1) Canadian Cool temperate 165.4 MgC/ha (1) Canadian Moderate temperate 189.9 MgC/ha (1) Canadian Cordilleran 185.5 MgC/ha (1) Canadian Interior cordilleran 236.4 MgC/ha (1) Canadian Pacific Cordillera 231.8 MgC/ha (1) Canadian Sub-arctic cordillera 223.8 MgC/ha (1) Young(<80 years old), Oregon and California Coast Range 31(27) MgC/ha(4) Young(<80 years old), Western Cascades 31(31) MgC/ha(4) Young(<80 years old), Sierra Nevada 8(8) MgC/ha(4) Young(<80 years old), Klamath Mountains 15(14) MgC/ha(4) Doug Fir and Hemlock Forest 259 - 274 MgC/ha(20, 34, 35, 36, 37, 48, 49) Canada - general - Medium aged 212.4 MgC/ha Canadian Boreal west 170.7 MgC/ha Canadian Boreal east 164.8 MgC/ha Canadian sub-arctic 266.1 MgC/ha Canadian Cool temperate 220 MgC/ha Canadian Moderate temperate 339.8 MgC/ha Canadian Cordilleran 225.7 MgC/ha Canadian Interior Cordilleran 323.9 MgC/ha Canadian Pacific Cordillera 374.6 MgC/ha Canadian Sub-arctic cordillera 237.2 MgC/ha References Turner 1975 Turner 1975 Boone 1998 Cole 1968 Harmon 1990 Grier 1971 Harmon 1990 Harmon 1990 Turner 1975 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Hudiburg 2009 Harmon 1990, Grier 1971, Cole 1968, and Boone 1998 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Kurz and Apps 1999 Modeling carbon storage under forest restoration Because we do not have 150 years to try out various forest restoration approaches, we must rely on models to predict the path that restoration may take. My approach is to use the U.S. Forest service’s growth and yield model FVS (Crookston, 2005). Forest managers across the US use this model because it includes regional variants tuned to local forest types and is relatively simple to use. This model uses the FIA plots and the model is based on empirical relationships derived from these forest plots (Crookston, 2005). This approach is therefore reliable when there is data underlying its predictions. The restoration scenarios which I am going to test are most likely beyond the initial scope of the model. The model has some problem calculating shading and its effect on forests of mixed ages. In addition, the establishment function of the Pacific Northwest variety is not enabled, therefore recruitment was added manually. Despite the fact that we are pushing the model, experimental data exists which can be used to check the results of the models. The equations which underlie FVS may lead to an underestimation of carbon storage in soil and coarse woody debris and an overestimation of stem storage. FVS assumes that coarse and fine woody biomass decays as a pool. Yet much other work has demonstrated that the decay rate of individual woody fragments is negatively correlated with size and that decay proceeds independently on each piece of wood. Therefore, the growth and death of larger trees should increase carbon storage. FVS will tend to assume that these materials decay too fast because the pools are lumped and because the rates do not change with the diameter of the CWD. Soil carbon is not directly modeled, only root biomass is addressed. Root biomass is grown according to a standard set of allometric equations. Therefore, we would expect the model to predict a lower value of carbon storage than the literature would suggest is possible. Carbon Storage FVS Model Outputs In order to test whether FVS would make reasonable predictions, I ran the model in a no-management scenario for 200 years to see what it predicted in terms of overall biomass. Figure 1 show the results of this simulation showing biomass storage through time. Biomass increases most rapidly in young forests, but is still increasing when the forest is 250 years old. At the end of the simulation, the forest had stored 1000 MgC/ha which falls into the upper end of the estimates from the literature. This estimate is higher than Luyssart’s maximum estimate of 700 MgC/ha (Luyssaert et al., 2008) but smaller than Smithwicks estimate of from other old growth forests (Smithwick et al., 2002). Compared to Luyssart’s estimate, this forest is storing twice as much carbon in its stems, but much less in CWD and roots than other studies would suggest. Forest carbon under a 35 different management scenarios was modeled with FVS. Despite a wide range of yearly thinning intensities (Average Proportion of trees cut each year of the simulation), carbon storage ranged from 811 MgC/ha at the lightest treatment to 577 MgC/ha At the heaviest treatment (Figure 18,Figure 19). Short return intervals resulted in lower carbon storage than longer return intervals. The 30 year to 50 year return intervals were almost identical for all thinning intensities (Figure 18,Figure 19). This suggests that the carbon consequences of these longer return intervals depend more on the proportion of trees cut than the thinning frequency. For the short return intervals, there appeared to be a carbon floor where more intense thinning did not result in less carbon storage. When treatments were more frequent and of higher intensity, the return interval was a better predictor of carbon than the intensity of the treatment. In order to understand the forests response to all treatments, I calculated the yearly treatment intensity, which was the proportion of trees cut divided by the return interval (Figure 18). When compared to the carbon contained in the stand, it shows an exponential decay from no management to 2% of trees cut per year and then a flat response at out to 5% cut per year (Figure 19). To give these findings more context, the thinning results were compared to a no management and a rotation forestry management option (Figure 21). The carbon storage under most of the restoration scenarios is much closer to the no management carbon storage than the active management carbon storage. Active management storage are much lower because clearcutting removes carbon and accelerates decomposition. It appears that increasing structural diversity and carbon storage are not mutually exclusive, and could both be pursued by future forest restoration projects. Carbon storage in western US ecoregions 400 Stored Carbon (MgC/ha) 350 300 250 200 Young 150 Mature 100 Old Growth 50 0 Oregon and California Coast Range Western Cascades Sierra Nevada Klamath Mountains Region Figure 16: Carbon storage across ecoregions of the western US, notice that the pacific forests, mainly composed of Sitka Spruce and Western hemlock, store the most carbon, and contain the greatest difference in storage between young and old forests. Figure 17: A comparison of carbon storage in young and old growth Sitka Spruce Forests compiled from the literature cited in table 3 and 4 Carbon Storage (MgC/ha) Thinning intensity versus total carbon storage 900 850 800 750 700 650 600 550 500 0 0.01 0.02 0.03 0.04 0.05 0.06 Thinning intensity Figure 18: Thinning intensity (percent of trees removed per year) versus carbon storage at age 150. Notice that carbon storage is sensitive to low thinning intensities but ending carbon storage values hit a floor after 2% removal per year. This is likely because heavy thinning is followed by more rapid growth due to higher light levels and decreased competition for remaining trees Thinning type versus carbon storage 900 Carbon storage (MgC/ha) 850 800 10 years 750 20 years 700 30 years 650 40 years 600 50 years 550 No Thinning 500 0 0.1 0.2 0.3 0.4 0.5 Proportion of overstory trees cut Figure 19: Carbon storage versus the proportion of trees cut per treatment. The colored lines correspond to the frequency of thinning treatments. Notice that thinning every 30 – 50 years results in the same ending carbon storage. Figure 20: Predicted Carbon storage for a moderate thinning (15% removal every 30 years) compared with the carbon storage in a young and Sitka Spruce old forests Carbon storage under three management regimes 1000 Carbon Storage (MgC/ha) 900 800 700 600 No management Restoration (15% removal every 30 years) Timber management 500 400 300 200 100 0 2013 2033 2053 2073 2093 Year Figure 21: Carbon storage under a moderate restoration thinning compared to a no-management and a timber management scenario Carbon Storage Forest restoration reduces the carbon storage in maturing forests below that of a no management approach. However, only a minor of loss of storage capacity will occur, 10% to 15% of total carbon storage possible under a no management approach. Yet even this slightly decreased carbon storage will still contain much more carbon than a young even aged forest. The results of the FVS model may overestimate the carbon stored in tree boles, and probably underestimates carbon storage in coarse woody debris and soil. Yet these errors are unlikely to change the conclusion that forest restoration is a successful way to both restore the ecosystem function of an old growth stand while increasing carbon storage. These two goals are not mutually exclusive and forest restoration may be in interesting way to align these interests to help improve forests and the global climate. Next Steps The carbon analysis performed by FVS is somewhat simplistic. Soil carbon storage is not addressed strongly in FVS. In addition, the decay functions for dead organic matter lead to faster than normal decay rates, especially for large coarse woody debris. Therefore, it would be helpful to take the FVS outputs and run one’s own analysis of carbon storage on the components. Another improvement would be to use the methodology used in the California Air Resources Board carbon trading program. This would allow a comparison of carbon storage to determine if carbon credits could be monetized from a restoration project to finance restoration activities.