Converting Even-Aged Plantations to Uneven-Aged Stand Conditions: a Simulation Analysis of Silvicultural Regimes with Slash Pine (Pinus Elliottii Engelm.)

Home , High grading

Converting Even-Aged Plantations to Uneven-Aged Stand Conditions: A Simulation Analysis of Silvicultural Regimes with Slash Pine (Pinus elliottii Engelm.)

Ajay Sharma, Kimberly Bohn, Shibu Jose, and Wendell P. Cropper

There has been increasing interest in managing forest stands as uneven-aged structures to promote sustainable harvests as well as maintain ecosystem services. This study provides a framework for simulating conversion of mature even-aged stands to uneven-aged slash pine (Pinus elliottii Engelm.) stands using the USDA Forest Vegetation Simulator (FVS) model. A total of 73 scenarios, representing combinations of two harvest methods (based on either “BDq” and/or “low thinning”), two harvesting cycles (10 or 20 years), three harvest intensities (4.6 or 8.0 or 11.5 m2 haϪ1 residual basal areas), and six levels of regeneration (0–2,224 seedlings haϪ1) were evaluated for structural diversity, timber production, and carbon (C) stocks over a 100-year period. The BDq harvest approach, which applied selection cutting based on diameter regulation from the first cutting cycle onwards, resulted in higher structural diversity. Scenarios based on low thinning in the first cutting cycle and BDq method from the third cutting cycle onwards tended to result in higher total merchantable timber and C stocks over the entire simulation period, particularly at higher residual basal areas and longer cutting cycles. None of the scenarios maximized all of the three variables simultaneously. Based on the desired objectives, land managers can choose among scenarios presented. The study revealed that regeneration and establishment of as low as 247 seedlings haϪ1 can lead to successful conversion and multiple benefits from uneven-aged slash pine stands. Keywords: stand conversion, forest vegetation simulator, structural diversity, volume production, flatwoods

neven-aged stands are structurally diverse with three or primary objective of management or even the most profitable prod- more age classes and exhibit, in a balanced state, a reverse uct or service produced (Tarp et al. 2005, Pukkala et al. 2010, UJ-shaped dbh distribution (Smith et al. 1997, Nyland Pukkala et al. 2011, Laiho et al. 2011). Biodiversity, wildlife habitat, 2002). Such stand conditions are often viewed as ecologically and recreation, and environmental services such as carbon sequestration economically efficient systems for managing forestlands (O’Hara are becoming increasingly important (Sedjo 2001, McConnell and Nagel 2006, Tahvonen et al. 2010, Boncina 2011, Kuuluvainen 2002, Food and Agriculture Organization [FAO] 2006, Gorte et al. 2012), and it is widely believed that uneven-aged silviculture 2009) and may be incompatible with the forest structure associated can improve forest resource use for multiple objectives in a sustain- with even-aged stands (MacArthur and MacArthur 1961, Thomas able manner (Brockway et al. 2005, O’Hara and Nagel 2006). Al- 1979, McConnell 2002, Laiho et al. 2011, Marion et al. 2011). though from a timber and fiber production standpoint even-aged Uneven-aged forest stands inherently may have higher resilience and systems have long been generally perceived to be more efficient and potential to adapt to changing climate (Guldin 2011) and, thus, profitable (Cafferata and Kemperer 2000, Henderson 2008, have been suggested to be an appropriate form of future risk man- Axelsson and Angelstam 2011), these products are not the only agement. Uneven-aged silviculture is also believed to mimic natural value derived from forestland, and increasingly, they are not the disturbances in sustaining many forest ecosystems (Brockway et al.

Manuscript received June 29, 2013; accepted March 6, 2014; published online April 3, 2014. Affiliations: Ajay Sharma (ajay.sharma@ufl.edu), University of Florida, West Florida Research and Education Center, Milton, FL. Kimberly Bohn (kkbohn@ufl.edu), University of Florida, Milton, FL. Shibu Jose ([email protected]), University of Missouri, Columbia, MO. Wendell P. Cropper (wcropper@ufl.edu), University of Florida, Gainesville, FL. Acknowledgments: The study was supported by the cooperative for Conserved Forest Ecosystems–Outreach and Research (CFEOR) funding, McIntire Stennis, and the University of Florida, School of Forest Resources and Conservation. The authors would like to express their sincere gratitude to Michael Morgan, Melissa Kreye, Justin McKeithen, and others who helped in the fieldwork and the Florida Forest Service for the opportunity to use the state forestlands and for in-kind support in timber marking and facilitating logging operations.

Forest Science • MONTH 2014 1 2005, Sharma et al. 2012). Not only are the uneven-aged forest (Hanewinkel and Pretzsch 2000) and PINEA IRR for stone pine stands valued for their ecosystem services (Axelsson and Angelstam (Pinus pinea L.) (Calama et al. 2008). The FVS model, unlike spe- 2011, Boncina 2011), but many comprehensive studies in a variety cies-specific models, has the ability to simulate conversion of a vari- of forest types have emphasized the equity or superiority of uneven- ety of species and forest types. In the present study, slash pine (Pinus aged stands in economic production as well (Haight 1987, O’Hara elliottii Engelm.), one of the commercially most important species and Nagel 2006, Tahvonen et al. 2010, Pukkala et al. 2011, Kuu- in southeastern United States, was used as a case study of this ap- luvainen et al. 2012). Changing public preferences and perceptions proach. Slash pine’s total acreage—including mixed stands with have led to increasing disapproval of even-aged methods such as longleaf pine (Pinus palustris Mill)—in the southeastern United clearcutting (Gobster 1996, Lindhagen 1996, Bliss 2000, Gunder- States was 5.3 million ha in 2007 (Smith et al. 2009). About 69% of sen and Frivold 2008), and the stand conditions associated with slash pine stands exist as intensively managed plantations with rota- uneven-aged forest structures are being viewed as esthetically favor- tion ages of 30 years or less (Barnett and Sheffield 2005). Owing to able (Silvennoinen et al. 2001, Bradley and Kearney 2007, Edwards slash pine’s strong shade-intolerant nature (Baker 1949, Lohrey and et al. 2012). Any form of forest management must consider all the Kossuth 1990), the transition to an uneven-aged stand would pos- diverse benefits, including timber production and provision of eco- sibly require a low residual basal area to promote an adequate system services, ranging from biodiversity to carbon storage to var- amount of understory light for regeneration and seedling growth. A ious social functions (FAO 2006, Diaci et al. 2011). Ϫ basal area approaching as low as about 4 to 5 m2 ha 1, as would be As a consequence, there has been a renewed and increasing inter- used with the irregular shelterwood method, has been suggested for est in managing forest stands as uneven-aged structures (Nyland 2003, Pommerening and Murphy 2004, Loewenstein 2005, Mizu- slash pine conversion and management, and such scenarios are cur- naga et al. 2010, Diaci et al. 2011). Many agencies and some private rently being evaluated in field trials on some Florida state forestlands landowners now have a mission that includes conversion of planta- (Brockway and Outcalt 2010, Sharma et al. 2012). This approach tions into uneven-aged stands (Guldin and Farrar 2002, Loewen- would likely result in adequate natural regeneration and, in partic- stein and Guldin 2004, Florida Division of Forestry 2007, Brock- ular, would help maintain a species-rich understory and open pine- way and Outcalt 2010). However, our experience with such savanna conditions (Jose et al. 2006, Brockway and Outcalt 2010). 2 Ϫ1 conversion and uneven-aged silviculture is limited. Since most of However, a residual basal area of up to 11.5 m ha has also been the research and practice in the past century has focused on even- reported to have no adverse effect on the early seedling establish- aged silviculture, forest managers lack information to guide their ment or growth of slash pine (McMinn 1981, Brockway and Out- long-term stand conversion projects. Additionally, the implications calt 2010) and may, thus, be one of the possible levels of appropriate of stand conversion and application of uneven-aged silviculture in residual basal areas to maintain in uneven-aged stands for timber managing forestlands for multiple benefits are uncertain. Managing production and ecosystem services. Intermediate levels of residual Ϫ forests for balanced production of wood and providing ecosystem basal area (e.g., 8 m2 ha 1) are also options for uneven-aged man- Ϫ services on a sustainable basis is critical for meeting the demands of agement of pine stands in Florida. Overall, 4.6 and 11.5 m2 ha 1 the 21st century (United Nations 2007, Diaci et al. 2011). may be considered as two extremes of residual basal area levels Several strategies have been suggested to convert and subse- within which slash pine can be managed as uneven-aged stands for quently manage plantations as uneven-aged stands. These typically multiple benefits. The lower level may be more conducive to pri- involve partial cuttings either uniformly over the stand or in patches, marily promoting open pine-savanna conditions with diverse intensive enough to allow successful establishment of multiage co- groundcover dominated by grasses and forbs while the higher basal horts and eventually allowing the application of uneven-aged silvi- area would be more reflective of a stronger objective of timber pro- cultural systems such as single tree selection or group selection sys- duction along with other ecosystem services. tem (O’Hara 2001, Guldin and Farrar 2002, Nyland 2003, The success of uneven-aged silvicultural methods also depends Loewenstein 2005, Kerr et al. 2010). One approach to partial cut- on its ability to promote regeneration of desired species and to have ting of even-aged stands involves heavily thinning the even-aged that regeneration develop into merchantable size classes (Shelton stand in the beginning by removing the inferior or suppressed trees and Cain 2000). However, our understanding about the minimum to create a vigorous residual stand to which a new age class will be level of regeneration that would be periodically required for success- added following natural regeneration (Nyland 2003, Loewenstein ful application of uneven-aged silviculture still remains limited 2005). Another suggested approach is to apply the selection cut (Guldin 2006). Since slash pine in the southeastern United States based on diameter regulation, which aims at creating a reverse-J shaped diameter distribution directly (Della-Bianca and Beck has been mostly cultivated as intensively managed, high-density, 1985). Field trials to evaluate these strategies generally span a period short-rotation plantations for the purpose of timber and fiber, sci- of many decades before their relative effectiveness can be evaluated. entific literature is critically deficient in studies on its natural regen- Simulation models, however, can be used to predict stand dynamics eration. This is, perhaps, true for a majority of the species that have or to conduct sensitivity analyses on different silvicultural regimes in long been artificially regenerated for high-intensity plantations. lieu of evidence that will take several decades to develop with exper- Lacking long-term studies of natural regeneration, simulation mod- imental field trials (Vanclay 1994, Weiskittel et al. 2011). eling is a useful starting point for addressing this question. We In this study, we used USDA Forest Service’s Forest Vegetation simulated stand conversion over a range of regeneration possibilities Simulator (FVS) model (Crookston and Dixon 2005) to evaluate a to identify optimum levels of regeneration that would satisfy forest variety of silvicultural regimes to convert a mature plantation to an management objectives. Because regeneration in forest ecosystems is uneven-aged stand. A few other studies have used species-specific influenced by a number of factors related to management, including models to simulate uneven-aged stand growth or conversions such residual basal area, site preparation, competition control, and other as the SILVA model for Norway spruce (Picea abies L. Karst.) species-specific disturbances such as prescribed fire and herbivory,

2 Forest Science • MONTH 2014 (Dickens et al. 2004, Jose et al. 2006), an appropriate level of regen- that adjusts the internal growth models to match the measured eration, when known, can be targeted by manipulating manage- growth rates of a particular location consistent with the source of the ment practices and/or supplementing natural regeneration with input data. Thus, it is able to modify predictions for local condi- planting. tions. The growth cycles in the simulator are set to 5- or 10-year time Thus, the overall objective of the study was to evaluate different steps with up to 40 cycles allowed per simulation (Dixon 2002, silvicultural regimes, over a range of levels of regeneration, for their Crookston and Dixon 2005). feasibility to successfully convert plantations to sustainable uneven- The model has the ability to simulate many types of harvest and aged stands. The FVS model used in the study is a highly versatile other silvicultural activities. The user has the options to specify how system that is capable of modeling a variety of forest types and stand many trees or the basal area from which size classes should be re- structures ranging from even-aged to uneven-aged by specifying any tained at the end of any simulation cycle along with other activities combination of management activities during any cycle of the sim- such as shrub removal or artificial planting. Typical growth and ulation period (Crookston and Dixon 2005). We evaluated two mortality rates are predicted as functions of postharvest conditions harvest approaches for stand conversion at three harvest intensities (residual densities/basal area etc.). Broadly, the FVS model has four and two cutting cycles. The regimes were assessed in terms of their primary components: diameter growth, height growth, mortality, long-term ecological and economic benefits. Specifically, we evalu- and crown change. Diameter growth, height growth, and crown ated stand structural diversity, carbon (C) stocks, and merchantable change models are further divided into submodels that pertain to timber production over a 100-year period. “small” trees (Յ 7.6 cm dbh) and those pertain to “large” trees (Ͼ 7.6 cm dbh). The small tree growth relationships are driven by Materials and Methods tree height from which diameter growth is then estimated. On the Simulation Input Data other hand, the large tree growth relationships are driven by diam- We used data collected from a mature slash pine plantation at eter first, and then height growth is estimated from diameter growth Tate’s Hell State Forest, Florida, to initiate simulation of stand and other tree and stand variables. The FVS-sn model is a widely conversion. Tate’s Hell State Forest (29.83N, 84.79W) is a poorly used model (Teck et al. 1996, Gilmore 2003, Johnson et al. 2007, drained, lowland, mesic-hydric flatwoods site between the Apalachi- Sorensen 2011, Saunders and Arseneault 2013) and more details cola and Ochlockonee Rivers in the panhandle of Florida. Large- about its contents, description, structure, and applications can be scale silvicultural operations and hydrologic manipulations in Tate’s found in McGaughey (1997), Van Dyck and Smith (2000), Don- Hell State Forest during the 1960s through the 1980s converted nelly et al. (2001), Dixon (2002), Crookston and Dixon (2005), extensive areas of native habitats to slash pine plantation. Stands and Keyser (2008). were established following intensive mechanical site preparation, FVS-sn does not have a full regeneration model. The user must bedding, and planting at high densities. Fertilizers, particularly ni- schedule natural regeneration events by specifying input values for trogen and phosphorus, were applied at midrotation. These large- density and size of expected new trees based on other published scale disturbances and manipulations to the habitats altered the literature and/or its understanding of species ecology. However, it ecological communities and hydrology of the area (Gilpin and Vow- can simulate regeneration from sprouting species, the feature we ell 2006). Currently, one of the main land management goals in turned off in our simulations. Tate’s Hell State Forest is to convert and maintain these plantations The inventory data collected from the slash pine plantation were as low-density uneven-aged stand structures that will maintain bio- used to initialize the simulations in FVS model using specific control logical diversity, while integrating public use (Florida Division of variables (Table 1; Figure 1). Parameter inputs (e.g., location, eco- Forestry 2007). logical codes, etc.) were selected that best represented the site con- The input data were collected from a mature unthinned slash ditions at Tate’s Hell State Forest. The outputs generated from FVS pine plantation scheduled for active conversion to an uneven-aged were used to calculate structural diversity, carbon stocks, and mer- stand. For this, we established five sample plots of 25 ϫ 25 m in the chantable timber production as described below. stand. Within each plot, we recorded dbh and species for all the trees greater than 10 cm dbh. Within the 25 ϫ 25 m plot, we also Silvicultural Regimes established two 5 by 5 m plots at diagonally opposite corners in We followed two approaches in developing silvicultural regimes which we measured all trees smaller than 10 cm dbh. for stand conversion. One approach was the direct application of the uneven-aged BDq method beginning with the first cut. The BDq Model Description method is a common approach used in developing marking tallies We used the FVS model Version 4280—Southern US for single tree selection system where “B” refers to residual basal area, PROGNOSIS RV: 12/20/11, developed by USDA Forest Service. “D” is the maximum diameter class to retain in stand, and “q” is FVS is a distance-independent, individual-tree forest growth model diminution quotient. Specifications of B, D, and q constitute a that uses individual stands as the basic units of projection. The unique reverse J-shaped distribution of diameter, which, theoreti- applications of FVS span a wide variety of forest types and stand cally, characterizes an ideal balanced uneven-aged stand. The diam- structures ranging from even-aged to uneven-aged and single to eter distribution of the stand to be cut is compared to that of the mixed species. FVS can model single stands or landscapes and large ideal, balanced uneven-aged stand, and a marking tally is generated regional assessments involving thousands of stands. The key vari- that brings the existing stand closer to the balanced uneven-aged ables of each tree are its species, diameter, height, crown ratio, di- structure (Farrar 1996, Smith et al. 1997, O’Hara and Gersonde ameter growth, and height growth; however, only species and diam- 2004, Guldin 2006, Keyser 2008). In the FVS model, the BDq eter information is mandatorily required. The key variables for each method is implemented by supplying the specifications for B, D, sample point, or plot, include slope, aspect, elevation, density, and a and q along with the diameter class width and a cutting cycle. In measure of site potential. The model has a self-calibration feature cases where deficit diameter classes are found, excess basal area is

Forest Science • MONTH 2014 3 Table 1. Control variables inputs specified to FVS-sn variant to A total of 73 scenarios were simulated, each with three replica- control the simulation for the study. tions. Each scenario represented a unique silvicultural regime con- Parameter or attribute Input setting sisting of a harvest type, residual basal area, cutting cycle, and regeneration density (Table 2). These scenarios varied from “no action” to Location Code 8,0501 (Apalachicola National Forest in Florida, Latitude: direct application of “BDq approach” beginning with the first cut 30.44 N, Longitude: 84.28W) (hereafter called “BDq scenarios”), to “low thinning in the first cut, Ecological Unit codes 232Dd (Gulf coastal lowlands) thinning across diameter classes at the second cutting cycle, and Slope (%) 0 BDq-based cut from third cutting cycle onwards” (hereafter called Aspect 0 Elevation(m) 9.15 “low thin scenarios,” named after the first cut). For all the cuts based Site Species Codes (Alpha code) SA, PC, MG,BY on BDq in both BDq scenarios and the later cuts in the low thin Site Index(m) for slash pine 27.43 scenarios, we set the values of parameters such that D Ͼ 86.4 cm and Maximum density (trees/ha) 1075 ϭ Number of projection cycles 21 q 1.5 for diameter-class width of 5 cm. The residual basal areas 2 Ϫ1 Projection cycle length 5 following harvests were either 4.6, 8.0, or 11.5 m ha (hereafter Volume equations National Volume Estimator called “low,” “intermediate,” or “high”) on a 10- or 20-year (here- Library after called “short” or “long”) cutting cycles during the simulation Pulpwood Sawtimber period. Regeneration scenarios varied from complete failure of nat- Volume Specifications ural regeneration (hereafter called “no regeneration scenarios”) to Ϫ Minimum dbh/top 10.2/10.2 25.4/17.8 abundant regeneration of as many as 2,224 seedlings ha 1 in incre- Diameter inside bark (cm) Ϫ1 Stump height (cm) 30.5 30.5 ments of about 495 seedlings ha during each cutting cycle. For Sampling Design these simulations, we specified regeneration as the seedlings of slash Ϫ Large trees (fixed-area plot) 6.5 pine with average height of 60 cm surviving 3 years after each cut- Small trees (fixed-area plot) 81 Breakpoint dbh (cm) 10.2 ting cycle. Each simulation was run for a total of 100 years after the first cut at simulation cycle lengths of 5 years. The naming scheme for the scenarios consisted of a sequence of four variables in the form “basal area-harvest type-cutting cycle-regeneration.” For example, the scenario “11.5-thin-10–247” represents a regime with residual Ϫ basal area of 11.5 m2 ha 1, treated with the low thinning harvest Ϫ method, a cutting cycle of 10 years, and 247 seedlings ha 1 of regeneration following every harvest. Model outputs included a tree list, a carbon report, a stand summary, and a cut list for each simulation (Dixon 2002). The individual trees in the tree list were then classified into diameter classes and height classes. We used a total of 18 diameter classes with class widths of 5.08 cm, ranging from size 0 (less than 2.54 cm) to 86.4 ϩ cm (equal to or greater than 83.8 cm). For height, we used a total of 16 height classes with class widths of 3.05 m, ranging from 0 (less than 1.52 m) to 45.72 ϩ (equal to or greater than 44.2 m). In both the cases, the classes represented the midpoints of the class widths.

Evaluation of Silvicultural Regimes Figure 1. FVS processing as used in simulating different silvicultural regimes for converting even-aged plantations to uneven-aged The scenarios were evaluated based on the resulting stand struc- stands. tural diversity, C stocks, and merchantable timber. Stand Structural Diversity proportionally distributed to the other classes to maintain the target At every cycle of the simulation period, we calculated the Shan- residual basal area (Keyser 2008). non Diversity Index for both the diameter class and height class The other conversion approach we used involved thinning from distributions with respect to the basal area constituted by them. The below during the first cutting cycle, followed by thinning across ϪΑ Shannon Index was calculated as equal to pi ln pi, where p is the diameter classes at the second cutting cycle, and then following with proportion of basal area constituted by a diameter class i (Magurran BDq based cuts from third cutting cycle onwards (Table 2). “Thin- 1988). Stand structural diversity was then obtained as the average of ning from below” or “low thinning,” involved cutting all the trees of the Shannon Index values for both diameter and height classes. small size first followed by larger-sized trees till a desired residual Every simulation had 21 values of Shannon indices for diameter and basal area was achieved. During the first cut this created a residual height distributions each, one for every cycle plus one at the begin- stand consisting of trees with large diameters. At the second cutting ning of the simulation. Values were plotted over time and the aver- cycle, thinning across all the diameter classes resulted in removal of age of Shannon Index values at all cycles during a simulation was some trees from both established age classes during this time period. calculated as average stand structural diversity under a given From the third cutting cycle onwards, a BDq cut similar to our other scenario. harvest regime was implemented across the three established cohorts following regeneration and growth modules of the simulator. Thus, C Stocks the two harvest approaches simulated in the study differed only at The outputs of the C reports obtained from the model were the first two cutting cycles. grouped into aboveground stored C (consisting of aboveground live

4 Forest Science • MONTH 2014 Table 2. Overview of the scenarios used in simulating conversion of slash pine plantations to uneven-aged stands. Each of the harvest types (BDq or low thin) was simulated over each combination of residual basal area (4.6 or 8.0 or 11.5 m2 ha؊1), cutting cycle (10 or yr), and level of regeneration (0 to 2224 seedlings ha؊1) as described below. All the scenarios had maximum diameter (D) > 86.4 20 .for simulating BDq cuts in ﬁnal stages 1.5 ؍ (cm, and diminution quotient (q

Harvest Residual basal areas Cutting cycles type Description of silvicultural regime (m2 haϪ1) (years) Regeneration ranges (seedlings/ha)a BDq BDq cut from first cutting cycle onwards 4.6, 8.0, 11.5 10, 20 0, 247, 741, 1,236, 1,730, and 2,224 Low thin 1. Thinning from below during first cutting cycle 4.6, 8.0, 11.5 10, 20 0, 247, 741, 1,236, 1,730, and 2,224 2. Thinning across diameter classes during second cutting cycle 3. BDq cut from third cutting cycle onwards No action No cut of any kind NA NA No harvest and no regeneration were simulated in this scenario. a The regeneration is defined as the seedlings of slash pine with average height of about 60 cm surviving 3 yr after each cutting cycle. The name of a scenario consisted of four parts representing respectively “basal area-harvest type-cutting cycle-regeneration.” tree, standing dead trees, down deadwood, forest floor, and the scenarios within the three levels of basal area, i.e., within the understory), below ground stored C (consisting of below ground groups of scenarios with 11.5 or 8.0 or 4.6 m2 ha-1 basal areas. live tree and below ground dead tree), and total C removed in Tukey’s Honestly Significant Difference (HSD) test was performed harvest at each year during the simulation period (Hoover and Re- at ␣ ϭ 0.05 to test for significant differences in all of the analyses. bain 2011). The total stand carbon at a given time was the sum of For each level of residual basal area, we then selected the top five above and below ground stored carbon at that time. The following scenarios based on the means that maximized stand structural diver- variables were calculated: sity, C stocks, and merchantable timber. Additionally, the top five (a) Total stand C in the beginning of simulation (year 0) ϭ Total scenarios leading to maximization of multiple benefits (i.e., collec- aboveground stored C in year 0 ϩ Total below ground stored C in tive provision of structural diversity, carbon storage, and merchant- year 0. able timber production) were identified for each level of residual (b) Total stand C at the end of simulation (year 100) ϭ Total basal area. For this purpose, we first scaled the values of all three aboveground stored C in year 100 ϩ Total below ground stored C criteria variables as percent of their maximum among all scenarios in year 100. and summed them for each scenario. Absent a good metric to di- (c) Total additional C stored during simulation period ϭ (Total rectly assess the multiple attributes with different units and different stand C at the end of simulation (year 100) ϩ sum of C harvested economic and noneconomic values, we assumed that higher sums during different cycles in simulation period) Ϫ Total stand C in the corresponded to greater ability to provide multiple benefits. beginning of simulation (year 0). Average annual C stock for each scenario was then obtained by Results dividing the total additional C stored by 100. Initial Stand Conditions Ϫ The overstory was dense (1,136 trees ha 1) with an average basal Ϫ Timber Production area and mean quadratic diameter of approximately 28.7 m2 ha 1 From the FVS summary output generated in simulations, we and 18 cm, respectively. The diameters at breast height of the initial calculated values of the following variables: slash pine plantation stand exhibited typical bell-shaped distribu- (a) Total merchantable timber produced during simulation pe- tion associated with even-aged stands (Figure 2A and B). Occasion- riod ϭ Total merchantable timber removed during simulation ϩ ally sweetbay (Magnolia virginiana L.) and pond cypress (Taxodium Total merchantable timber left standing after year 100 Ϫ Total ascendens Brongn.) were found in the overstory. Slash pine made merchantable timber in year 0. about 63% of tree density and 96% of stand basal area. The lower (b) Total sawtimber produced during simulation period ϭ Total diameter classes were dominated by sweetbay that had regenerated sawtimber removed during simulation ϩ Total sawtimber left under the dense canopy of slash pine. Natural regeneration of slash standing after year 100 Ϫ Total sawtimber in year 0. pine was lacking. Average annual productions of these variables were then obtained by dividing the totals by 100. Evaluation of Silvicultural Regimes None of the scenarios maximized all criteria variables simultane- Statistical Analyses ously. There were tradeoffs involved between structural diversity As recommended by Hamilton (1991) for the application of and C stocks, as well as merchantable timber production in different FVS, we ran three simulations for each scenario, reseeding the ran- scenarios. The scenarios that led to higher structural diversity gen- dom number generator, and thus, producing variation in the pro- erally resulted in lower C stocks and merchantable timber produc- jection results (Hong and Mladenoff 1999, Haefner 2005). A low tion and vice versa (Tables 3–5; Figure 3A–D). and stable variance across the simulations was observed. Means were Not surprisingly, either harvesting approach followed by com- then calculated for average annual values of structural diversity, C plete failure of regeneration (no-regeneration scenarios) was among stocks, merchantable timber, and sawtimber for each scenario. Anal- the worst in terms of structural diversity and C stocks, as well as yses of variance (analysis of variance [ANOVA]) were carried out for merchantable timber production. The no action scenario, wherein all the scenarios to test for overall main effects of harvest type (BDq no cutting and no regeneration occurred during the period of sim- and low thin), residual basal area, cutting cycle, regeneration level, ulation, also led to low values of these variables. Values for stand and their two-way interactions. ANOVA were done separately for structural diversity, C stocks, and merchantable timber production

Forest Science • MONTH 2014 5 in a signiﬁcant (P Ͻ 0.001) harvest type by cutting cycle interaction. Within each harvest type, scenarios with higher residual basal or with the shorter cutting cycle led to higher average structural diversity. The top scenarios that maximized average structural diversity at Ϫ the high basal area of 11.5 m2 ha 1 included BDq scenarios with a 10-year cutting cycle and regeneration varying between 247 Ϫ through 2,224 seedlings ha 1 (Shannon Index ϭ 2.03 to 2.06; Ϫ Table 3). At the intermediate residual basal area of 8.0 m2 ha 1, BDq scenarios with 10-year cutting cycle and regeneration varying Ϫ between 247 through 1,730 seedlings ha 1 again resulted in the highest average structural diversity (Shannon Index ϭ 2.03 to 2.05). Ϫ At the low basal area of 4.6 m2 ha 1, BDq scenarios” with 10-year cutting cycle and regeneration of either 247 through 1,236 seedlings Ϫ ha 1 led to the highest average structural diversity (Shannon In- dex ϭ 1.97 to 1.99) (Table 4). Although low thin scenarios led to overall low average stand structural diversities over the simulation period, they had achieved comparable levels by the end of simulation and sometimes by year 50 (Figure 4A). The low overall average structural diversity in low thin scenarios was highly inﬂuenced by initial cutting cycles during the simulation period.

C Stocks Annual average C stocks increased significantly (P Ͻ 0.001) and substantially, by as much as three times, when shifting from no Ϫ regeneration to 247 seedlings ha 1 (Figure 3B). Thereafter, significant increases were observed with increasing regeneration in a few scenarios, mostly with longer cutting cycles. The sensitivity of C stocks Figure 2. Input stand conditions inventoried in 2009. (A) Slash to an increase in regeneration was evident when basal area was low and pine (Pinus elliottii Engelm.). exhibited bell shaped diameter distri- cutting cycle was long. Also, in general, low thin scenarios were more bution typical of even-aged stands. However, due to dense stand sensitive to regeneration than BDq scenarios. For example, low thin Ϫ conditions and absence of any hardwood control measures, sweet- scenarios with a 20-year cutting cycle and 4.6 or 8.0 m2 ha 1 residual bay (Magnolia virginiana L) had grown densely as an understory basal areas showed the highest sensitivity to increases in regeneration, and was fast becoming a part of midstory, (B) More than 95% of basal area was constituted by slash pine. The stand basal area was such that their C stocks became comparable to those of the scenarios 2 Ϫ1 .(m2 ha؊1. with 11.5 m ha at high regeneration levels (Figure 3B 28.7 Generally across all regeneration scenarios, we found significant interactions between harvest type with cutting cycle (P Ͻ 0.001) under the no action scenario were overall similar or slightly superior and with residual basal area (P Ͻ 0.001) on average C stocks. Gen- to the harvest scenarios with no regeneration. erally, longer cutting cycles resulted in higher C stocks than shorter cycles; however, C stock values derived from BDq scenarios were less Stand Structural Diversity sensitive to changes in length of cutting cycle than low thin scenarios There were significant interaction effects of regeneration level (Figure 3B). This was more evident at the low residual basal than with harvest type, cutting cycle length, and residual basal area. Av- intermediate or high residual basal areas. The main effect of harvest erage structural diversity increased significantly (P Ͻ 0.001), from type on C stocks was not significant. 0.75 to 1.76 when scenarios shifted from no regeneration to 247 Evaluation of the top scenarios at basal areas of 11.5 as well as 4.6 Ϫ Ϫ seedlings ha 1, though the magnitude of the increase differed by m2 ha 1 indicated that annual average C stocks were highest (2.23 Ϫ Ϫ Ϫ harvest types and cutting cycles (Figure 3A). Any further increases and 1.92 metric ton ha 1 year 1 for 11.5 and 4.6 m2 ha 1, respec- Ϫ from 741 to 2,224 seedlings ha 1 did not significantly change struc- tively) in low thin scenarios with a 20-year cutting cycle and high Ϫ tural diversity except for a few scenarios, resulting in a slight but levels of regeneration at 1,730 and 2,224 seedlings ha 1. The BDq statistically significant (P Ͻ 0.001) decrease in structural diversity scenario with a 20-year cutting cycle and regeneration level of 2,224 Ϫ with increasing regeneration levels at the lower residual basal area seedlings ha 1 was also among the top scenarios (Table 3). At a basal Ϫ and longer cutting cycle. area of 8.0 m2 ha 1, the highest annual average C stocks (2.24 Ϫ Ϫ Generally across all regeneration scenarios, harvest type had the metric ton ha 1 year 1) were led by the low thin scenario at 20-year Ϫ biggest influence on stand structural diversity value, which ranged cutting cycle and high regeneration of 2,224 seedlings ha 1 (Table Ϫ from 1.4 to 2.06 (Figure 3A). BDq scenarios resulted in higher 4), similar to the maximum realized at 11.5 m2 ha 1. average structural diversity than the low thin scenarios within any Over the simulation period, total stand carbon fluctuated as in- given residual basal area and cutting cycle (P Ͻ 0.001). However, fluenced by periodic harvests at different cutting cycles during the values for BDq scenarios at a 20-year cutting cycle were not different simulation period (Figure 4B). At any given time, total stand carbon from those of a low thin scenario at a 10-year cutting cycle, resulting was proportional to the basal area, and greater fluctuations in C

6 Forest Science • MONTH 2014 Table 3. Estimates (mean) of structural diversity, carbon stocks, and timber production in top scenarios at 11.5 m2 ha؊1 basal area. Five top scenarios for each of the variable that maximized their value and ﬁve scenarios that maximized multiple beneﬁts were selected. Eight scenarios overlapped resulting in a total of 12 scenarios.

Average stand structural Annual average merchantable Annual average sawtimber diversity (Shannon Annual average C stocks wood production production Rank (provision of Scenario Index) (metric ton/ha/year) (m3/ha/year) (m3/ha/year) multiple benefits) 11.5-BDq-10–0247 (D) 2.05a 1.77e 5.77f 5.02d,e,f 15 11.5-BDq-10–0741 (D) 2.05a 1.85d,e 6.00d,e,f 5.07b,c,d,e,f 9 11.5-BDq-10–1236 (D,M) 2.05a 1.89c,d 6.11c,d,e 5.04c,d,e,f 3 11.5-BDq-10–1730 (D) 2.06a 1.87d 6.00d,e,f 4.88e,f 8 11.5-BDq-10–2224 (D) 2.03a 1.87d 5.88e,f 4.78f 12 11.5-BDq-20–1730 (C,M) 1.85b 2.10b 6.09d,e 5.30a,b,c,d 5 11.5-BDq-20–2224 (C,M) 1.83b 2.17a,b 6.22c,d 5.39a,b,c 1 11.5-Thin10–1730 (T) 1.83b 1.91c,d 6.26b,c,d 4.93e,f 13 11.5-Thin10–2224 (T) 1.82b 1.97c 6.39a,b,c 4.88e,f 10 11.5-Thin20–1236 (C,T) 1.63c 2.12b 6.54a,b 5.47a 11 11.5-Thin20–1730 (C,T,M) 1.63c,d 2.18a,b 6.58a 5.40a,b 4 11.5-Thin20–2224 (C,T,M) 1.60d 2.23a 6.57a 5.20a,b,c,d,e 2 The name of a scenario consisted of four parts representing respectively “basal area-harvest type-cutting cycle-regeneration.” D, C, and T represent one of the top scenarios in structural diversity, carbon stocks, and merchantable timber production. M represents the top scenarios with multiple benefits. The values with the same letter are not significantly different.

Table 4. Estimates (mean) of structural diversity, carbon stocks, and timber production in top scenarios at 8.0 m2 ha؊1 basal area. Five top scenarios for each of the variable that maximized their value and ﬁve scenarios that maximized multiple beneﬁts were selected. Nine scenarios overlapped resulting in a total of 11 scenarios.

Annual average stand Annual average C Annual average merchantable structural diversity stocks (metric wood production Annual average sawtimber Rank (provision of Scenario (Shannon Index) ton/ha/year) (m3/ha/year) production (m3/ha/year) multiple benefits) 8.0-BDq-10–0247 (D) 2.05a 1.48g 4.62e 3.99b 16 8.0-BDq-10–0741 (D) 2.04a 1.54f,g 4.79d,e 4.02b 15 8.0-BDq-10–1236 (D) 2.04a 1.62e,f 4.94c,d 4.05b 10 8.0-BDq-10–1730 (D) 2.03a,b 1.59e,f,g 4.90c,d 3.88b 13 8.0-BDq-10–2224 (D) 1.99b 1.67e 5.12c 4.00b 8 8.0-BDq-20–1236 (T,M) 1.82c 1.95d 5.77b 5.03a 4 8.0-BDq-20–1730 (C,M) 1.80c,d 1.97c,d 5.64b 4.87a 5 8.0-BDq-20–2224 (C,T,M) 1.76d 2.09b,c 5.75b 4.81a 2 8.0-Thin20–1236 (C,T) 1.60e 2.02b,c,d 6.06a 5.00a 6 8.0-Thin20–1730 (C,T,M) 1.56e,f 2.10b 6.21a 4.97a 3 8.0-Thin20–2224 (C,T,M) 1.53f 2.24a 6.29a 4.91a 1 The name of a scenario consisted of four parts representing respectively “basal area-harvest type-cutting cycle-regeneration, D, C, and T represent one of the top scenarios in structural diversity, carbon stocks, and merchantable timber production. M represents the top scenarios with multiple benefits. The values with the same letter are not significantly different.

Table 5. Estimates (mean) of structural diversity, carbon stocks, and timber production in top scenarios at 4.6 m2 ha؊1 basal area. Five top scenarios for each of the variable that maximized their value and ﬁve scenarios that maximized multiple beneﬁts were selected. Nine scenarios overlapped resulting in a total of 11 scenarios.

Annual average stand Annual average C Annual average merchantable structural diversity stocks (metric wood production (m3/ha/ Annual average sawtimber Rank (provision of Scenario (Shannon Index) ton/ha/year) year) production (m3/ha/year) multiple benefits) 4.6-BDq-10–0247 (D) 1.99a 1.13e 3.23e 2.79b 17 4.6-BDq-10–0741 (D) 1.97a,b 1.16e 3.29d,e 2.68b 16 4.6-BDq-10–1236 (D) 1.97a 1.21d,e 3.45d,e 2.75b 13 4.6-BDq-10–1730 (D) 1.93b 1.26d 3.56d 2.86b 11 4.6-BDq-10–2224 (D) 1.93b 1.29d 3.59d 2.79b 9 4.6-BDq-20–1236 (M) 1.70c 1.70c 4.86c 4.05a 5 4.6-BDq-20–1730 (C,T,M) 1.62d 1.79b 5.03b,c 4.04a 3 4.6-BDq-20–2224 (C,T,M) 1.58d 1.84a,b 4.95c 3.82a 4 4.6-Thin20–1236 (C,T) 1.49e 1.75b,c 5.11a,b,c 4.14a 6 4.6-Thin20–1730 (C,T,M) 1.45f 1.89a 5.40a 4.19a 1 4.6-Thin20–2224 (C,T,M) 1.40g 1.92a 5.34a,b 3.99a 2 The name of a scenario consisted of four parts representing respectively “basal area-harvest type-cutting cycle-regeneration,” D, C, and T represent one of the top scenarios in structural diversity, carbon stocks, and merchantable timber production. M represents the top scenarios with multiple benefits. The values with the same letter are not significantly different.

Ϫ stocks were observed in scenarios with longer cutting cycles. For 106.14 metric ton ha 1 at cutting cycles of 10- and 20-year cutting example, after the ﬁrst cut, total stand carbon for low thin scenarios cycles, respectively. At low residual basal area, the low thin scenarios Ϫ Ϫ at 11.5 m2 ha 1 residual basal area with moderate regeneration level with moderate regeneration level of 1,236 seedlings ha 1 had total Ϫ of 1,236 seedlings ha 1 varied between 55.51 to 71.44 or 58.46 to stand carbon ﬂuctuating between 31.96 to 48.00 and 34.31 to

Forest Science • MONTH 2014 7 Figure 3. (A) Structural diversity (Shannon Index), (B) carbon stocks, (C) total merchantable timber, and (D) sawtimber production under different scenarios of harvest type (BDq or low thin, distinguished by solid and dashed lines), residual basal area (4.6 or 8.0 or 11.5 m2 ha؊1, distinguished by dark and faded lines), and cutting cycle (10 or 20 years, distinguished by separate markers) as affected by level of regeneration in slash pine simulated over 100 years.

Ϫ 74.25 metric ton ha 1 at cutting cycles of 10 and 20 years, Annual average merchantable timber production during the sim- respectively. ulation period followed somewhat similar patterns as annual average C stocks, with significant interactions of residual basal area and Ͻ Ͻ Merchantable Timber Production harvest type (P 0.001) and with cutting cycle (P 0.001). How- As with other response variables, the greatest increase in annual ever, there were no significant interactions between harvest type and Ϫ average merchantable timber production (as much as 4.2 m3 ha 1 cutting cycle. The scenarios with higher residual basal area had year Ϫ1) was observed when the scenarios shifted from no regener- significantly higher total merchantable timber production than the Ϫ ation to 247 seedlings ha 1 (P Ͻ 0.001). Further, slight but signif- scenarios with low basal area (P Ͻ 0.001); however, longer cutting icant increases were observed in many scenarios with increasing cycles did compensate for lower residual basal area, resulting in Ϫ levels of regeneration up to 1,236 seedlings ha 1 or higher, though comparable amounts of merchantable timber production as those of in some scenarios (mostly BDq scenarios at high and intermediate higher basal areas. The main effect of harvest type on total mer- residual basal areas) increases in regeneration over 1,236 seedlings chantable timber was not significant. Ϫ ha 1 led to decreases in total merchantable timber production. Sim- Interestingly, sawtimber production revealed very clearly the ef- ilar to carbon stocks, a greater sensitivity to increases in regeneration fect of residual basal area and cutting cycle (Figure 3D). The scenar- was observed in the low thin scenario with low residual basal area ios with longer cutting cycle had significantly (P Ͻ 0.001) and and a long cutting cycle (Figure 3C). substantially higher annual average sawtimber production than the

8 Forest Science • MONTH 2014 Figure 4. Changes in (A) stand structural diversity (Shannon index), (B) total stand carbon, and removal (yield) of (C) total merchantable timber and (D) sawtimber at multiple cutting cycles during the simulation period of 100 years in slash pine for scenarios with moderate level of regeneration (1,236 seedlings/ha) at residual basal areas of 11.5, 8.0, and 4.6 m2 ha؊1. Notably, low thin scenarios (dashed lines) approached to structural diversity values comparable to BDq scenarios at around year 50 as simulation progressed.

Forest Science • MONTH 2014 9 corresponding scenarios with shorter cutting cycle; the effect being 2009), and high structural diversity was achieved quickly in those more pronounced at lower basal areas. Sawtimber production also scenarios. In applying BDq to a mature even-aged stand in the Ϫ increased significantly (as much as 3.4 m3 ha 1 year Ϫ1) when the initiation of conversion, however, this method also tends to cut a Ϫ scenarios shifted from no regeneration to 247 seedlings ha 1 for all large proportion of mid-sized and several large-sized trees in the scenarios (P Ͻ 0.001). stand, leaving a residual stand of mostly inferior, suppressed trees The top scenarios resulting in maximum average annual mer- and a scattering of large trees (Kelty et al. 2003, Loewenstein 2005). chantable timber at the high residual basal area included low thin Notably, although the low thin scenarios resulted in low average scenarios with a short cutting cycle and high regeneration of 2,224 structural diversity, these approached similar structural diversity as Ϫ seedlings ha 1 and scenarios with a long cutting cycle but moderate the BDq scenarios by about year 50 as the simulation progressed. Ϫ to high regeneration of 1,236–2,224 seedlings ha 1 resulted in The low structural diversity in low thin scenarios in the beginning maximum average annual merchantable timber production of 6.39 was due to the fact that lower diameter classes were selectively re- Ϫ Ϫ to 6.58 m3 ha 1 year 1 (Table 3). At low and intermediate residual moved during that cutting cycle leading to concentration of large basal areas, the highest average annual merchantable timber sized trees, which reduced overall average structural diversity. How- Ϫ Ϫ production (5.11–5.40 and 6.06–6.29 m3 ha 1 year 1, respec- ever, this reduction was restricted to the first cutting cycle only and tively) were achieved in the low thin scenarios at 20-year cutting thereafter structural diversity increased rapidly to become compara- Ϫ cycle with regeneration between 1,236 to 2,224 seedlings ha 1 (Ta- ble to BDq scenarios. bles 4 and 5). Although uneven-aged stands with higher structural diversity As with total carbon stocks, total merchantable as well as saw- than the even-aged stands are recommended for improving wildlife timber production across the simulation period fluctuated between habitat in pine forests of southeastern United States (James et al. cutting and growth periods (Figure 4C and D). At a moderate level 2001, McConnell 2002, Marion et al. 2011), there is no reported Ϫ of regeneration of 1,236 seedlings ha 1, the total merchantable level of structural diversity that would be optimum for all relevant Ϫ timber yield varied from 28.5 to 74.5 m3 ha 1 and 79 to 156.2 m3 species. It is possible that the optimum structural diversity required Ϫ ha 1—for all levels of residual basal areas and harvest types—on to maintain habitat as well as aesthetics is not necessarily the maxi- cutting cycles of 10 and 20 years, respectively. Not surprisingly, the mum possible that could be achieved in uneven-aged stands. For lowest yields per cutting cycle were obtained at the lowest basal area example, red-cockaded woodpecker (Picoides borealis), a US feder- Ϫ of 4.6 m2 ha 1 at short cutting cycles. Values varied from 28.5 to ally listed endangered species, requires habitat characterized by open Ϫ 37.9 and 30.7 to 41.3 m3 ha 1 for BDq and low thin scenarios. For stand conditions consisting of large old pines (for cavity) and low the most part, total merchantable and sawtimber yields begin to density of small pines with little or no midstory and abundant native stabilize by the latter half of the simulation period. This is more bunchgrass and forb groundcover (US Fish and Wildlife Service evident for the shorter cutting cycle, perhaps because there have 2003). Also, structural diversity as defined in our study is essentially been more cuts to regulate structure than with the 20-year cutting a measure of diversity in tree diameters and heights in a stand. cycle within the 100-year time frame of the simulation. Structural diversity, in its most comprehensive sense, will also include snags, litter accumulation, coarse woody debris, and the spatial distribution of individual trees or age classes across the landscape Best Scenarios for Multiple Benefits which all together creates a complex structure (McElhinny et al. Most of the scenarios that maximized multiple objectives were 2005). These additional attributes should be examined in future similar across all residual basal area levels. For example, both BDq studies. and low thin scenarios with a 20-year cutting cycle and regeneration Ϫ The scenarios that led to high C stocks, in general, also resulted varying from 1,730 to 2,224 seedlings ha 1maximized multiple in high merchantable and sawtimber production. While other fac- benefits across all residual basal areas (Tables 3–5). Additionally, the Ϫ tors such as stocking density and forest productivity determine the BDq scenario with 20-year cutting cycle at 1,236 seedlings ha 1 also Ϫ amount of C stored in the forest biomass, the length of the rotation maximized multiple benefits at 4.6 and 8.0 m2 ha 1. At the high determines how long the C remains in the forest system (Kaipainen residual basal area, a BDq scenario with 10-year cutting cycle also et al. 2004). This is consistent with our model simulations as low was among the top scenarios at moderate level of regeneration of Ϫ thinning with a long cutting cycle maximized C stocks by retaining 1,236 seedlings ha 1 (Table 3). It may be noted that higher ranking larger trees, which added more C than the respective BDq scenarios of BDq scenarios was due to relatively greater contribution of struc- with similar cutting cycles. For a landowner who is more interested tural diversity. In other words, BDq scenarios were more effective at in direct economic benefits by selling timber and earning C credits, creating structural diversity as compared to timber production or C these scenarios are highly desirable. Although long-term empirical storage. For example, at the high basal area, average structural diver- data for C storage and timber production in slash pine managed as sity ranged from 1.83 to 2.05 in top Bdq scenarios while it was 1.60 uneven-aged stands is not available, other studies on uneven-aged to 1.63 in the top low thin scenarios. loblolly- shortleaf pines in the southern United States have exten- sively reported on timber volume productions (Reynolds 1969, Discussion Reynolds et al. 1984, Farrar et al. 1989, Murphy et al. 1991, Guldin Comparison of Silvicultural Regimes 2002). In uneven-aged mixed loblolly-shortleaf pine stands in the The scenarios that led to highest average structural diversity over Upper West Gulf Coastal Plain west of the Mississippi River, the simulation period at all levels of residual basal areas invariably Guldin (2002) reported that after 60 years, long-term annual vol- Ϫ involved BDq cut from the beginning. This is because the BDq ume production was 6–7.4 m3 ha 1 total merchantable volume and Ϫ approach specifically aims at creating a residual stand closer to the 5.0–5.5 m3 ha 1 sawtimber when the residual basal area was 11–13 Ϫ Ϫ balanced uneven-aged structure with a typical reverse J-shaped di- m3 ha 1. Our simulation outputs at 11.5 and 8.0 m2 ha 1 residual ameter distribution (Smith et al. 1997, Guldin 2006, Johnson et al. basal areas appear to be in agreement with these reported values.

10 Forest Science • MONTH 2014 Ϫ However, at the lower basal area of 4.6 m2 ha 1 (which was approx- At low and intermediate residual basal areas, either BDq or low imately 35–42% of basal area in treatment reported in the same thin scenario with a longer cutting cycle and moderate to high study by Guldin 2002), our simulation outputs are approximately regeneration will maximize the provision of multiple benefits. The 50% of total merchantable timber and 60% of sawtimber produc- top scenarios involving low thinning in the beginning will retain the tion of the values reported by Guldin (2002). best grown trees (dominants and codominants) in the residual stand Although the yields of total merchantable and sawtimber were (Smith et al. 1997, Nyland 2003) and may be advantageous. Such a relatively stable during the simulation period, particularly for low-thinned stand illuminates the understory by removing low shorter cutting cycles, the operability of the harvests in some of the shade (Nyland 2003). The vigorous residual stand with a bright scenarios at the lowest residual basal area may be an economic con- understory seems more likely to result in successful natural regener- Ϫ cern for some forest managers. A harvest of more than 25 m3 ha 1 ation and groundcover diversity. In that sense, low thin scenarios Ϫ sawtimber (Guldin 2002) or more than 31.5 to 52 m3 ha 1 have greater ecological feasibility than the BDq scenarios. pulpwood/chip and sawtimber (Huang et al. 2005, Eric Howell, The levels of residual basal area simulated in the study may be Florida Forest Service, pers. comm., Dec. 6, 2013) is generally con- considered to represent alternative densities for uneven-aged man- sidered an operable harvest, and thus, most of the scenarios would agement in slash pine for multiple benefits. As the study has shown, Ϫ lead to this. Only with a few cutting cycles in scenarios at the lowest at these three levels (4.6, 8.0, and 11.5 m2 ha 1) of basal area, the basal area and short cutting cycles would be less than the operable broad patterns of structural diversity, carbon storage, and timber harvests will be obtained. Thus, from economic operability point of production are similar, with only the magnitude of the responses view, the low basal area management should aim for longer cutting differing. This suggests that our observations regarding the effect of cycles. cutting methods on optimizing multiple benefits may be applicable Most public agencies are interested in managing forests for larger at many residual basal area levels though more empirical evidence societal, environmental, and economic benefits. This requires strat- may be needed to support the actual quantitative projections of egies that optimize the simultaneous provision of multiple benefits. simulations, especially at the lower basal areas. Our optimization approach scales the values of all three criteria variables quantified in the study (structural diversity, C stocks, and timber production) as a percentage of their maximum among all Assumptions and Limitations of the Study scenarios and then ranks the scenarios for maximizing the provision Although the simulation analysis provided a broad evaluation of of these three benefits. The use of scaling for this purpose does the conversion process and of the tradeoffs involved in different assume that values of a variable are distributed uniformly or linearly conversion scenarios, there were some limitations that should be within its range and that each of the three criteria has equal value to addressed. The FVS model is not spatially explicit and, hence, could a landowner, however this provided a basic means for provision of all not appropriately be used to simulate stand conversion using group three criteria among the 73 scenarios that we evaluated. selection or patch harvest cuts, which are considered among some of The simulation analysis revealed that the combination of multi- the best means to convert stands of intolerant species (Kelty et al. ple benefits of structural diversity, C stocks, and timber production 2003, Nyland et al. 2003). Arseneault and Saunders (2012), and at different residual high basal areas were maximally provided by Saunders and Arseneault (2013) addressed this limitation by using a either BDq scenarios with moderate to high levels of regeneration resampling approach to simulate group selection, shelterwood, and Ϫ (1,236–2,224 seedlings ha 1) or low thin scenarios with high levels single tree selection systems in FVS. The lack of a spatial model also Ϫ of regeneration (1,730–2,224 seedlings ha 1). However, some con- precluded us from evaluating the spatial complexity of age and size cerns may remain regarding the feasibility of some of these top distributions of trees across the forest, which may be inherent in scenarios. For example, the BDq scenarios may not be able to ade- natural southern pine ecosystems (Jose et al. 2006.) quately provide a vigorous source of seed trees to produce even a Although the FVS simulation outputs are derived from methods moderate level of regeneration during the initial harvests. The BDq and computations of C accounting consistent with the Intergovern- cut applied to a mature stand in the beginning essentially leads to mental Panel on Climate Change (IPCC) Good Practice Guidance high grading by removing a large proportion of large, phenotypi- and US voluntary C accounting rules and guidelines (Hoover and cally superior trees and creating a residual stand consisting of sup- Rebain 2011), the carbon reporting in our study did not include C pressed inferior trees (Smith et al. 1997, Kelty et al. 2003, Nyland et fluxes such as management-related emissions (e.g., emissions related al. 2003). Such a stand with poorly developed crown structure is to decomposing residues, harvest equipment use, timber transpor- likely to have reduced growth rates and may not result in production tation, etc.) A complete C footprint analysis would include life-cycle of adequate seed crop following the cut. Owing to this reason, the analysis of all aspects of forest management, including emissions BDq approach has been considered inappropriate for converting from site preparation activities, prescribed fires, and emissions re- stand structures (Kelty et al. 2003, Loewenstein 2005). Supplemen- lated to production, transportation, and application of fertilizers tal regeneration following the first two cuts may be needed to ensure and other chemicals, if used (Hoover and Rebain 2011). The chang- success of these scenarios. However, once an uneven-aged structure ing climate will likely have significant effect on growth and yield of has been achieved, the BDq approach can be successfully applied for the stand and was not accounted for in the study. maintaining uneven-aged structure (Kelty et al. 2003, Loewenstein One of the most important limitations to this long-term simula- 2005, Guldin 2006, Brockway and Outcalt 2010). Alternatively, in tion was the lack of a natural regeneration module for slash pine in the low thin scenarios, the residual stand would consist of vigorous FVS. We, therefore, conducted a sensitivity analysis to evaluate a superior trees. However, a high level of regeneration in a stand with range of plausible regeneration scenarios. Natural regeneration and high basal area may also be difficult to achieve as the high light growth in southern pines are determined by numerous factors that requirements of the species may not be met in dense stand include both intrinsic (i.e., silvics) and external influences (Shelton conditions. and Cain 2000). In slash pine, fire and competition from understory

Forest Science • MONTH 2014 11 Ϫ shrub and hardwood vegetation are important influences on seedlings ha 1 during a cutting cycle may not be economically prohib- ling germination and recruitment. These factors and their interac- itive to a landowner. tions were not specifically simulated in our study, however, we did indirectly account for these factors by assessing a range of regenera- Literature Cited tion scenarios from complete failure of regeneration to abundant regeneration that would span the likely range of regeneration den- ARSENEAULT, J.E., AND M.R. SAUNDERS. 2012. Incorporating canopy-gap sities with and without these additional control measures. This induced growth responses into spatially implicit growth model projec- helped us identify the levels of regeneration that could meet the tions. Ecol. Model. 237–238:120–131. AXELSSON, R., AND P. ANGELSTAM. 2011. Uneven-aged forest manage- management objectives. Although most of the best scenarios were ment in boreal Sweden: Local forestry stakeholders’ perceptions of dif- found to require moderate to high levels of regeneration, the regen- Ϫ1 ferent sustainability dimensions. Forestry 84(5):567–579. eration level of as few as 247 seedlings ha also led to reasonable BAKER, F.S. 1949. A revised tolerance table. J. For. 47(3):179–181. benefits (especially in stand structural diversity and sawtimber pro- BARNETT, J.P., AND R.M. SHEFFIELD. 2005. Slash pine: Characteristics, duction at all basal areas) with either decrease or only marginal gains history, status, and trends. P. 1–6 in Slash pine: Still growing and grow- with increasing regeneration densities for many scenarios. ing! Proc. of the slash pine symposium, Dickens, E.D., J.P. Barnett, W.G. Land managers and practitioners can then target to achieve those Hubbard, and E.J. Jokela (eds.). USDA For. Serv., Gen. Tech. Rep. levels of regeneration by using information about the species and SRS-76, Asheville, NC. 145 p. manipulating management activities (Shelton and Cain 2000). For BLISS, J.C. 2000. Public perceptions of clearcutting. J. For. 98(12):4–9. example, seedbed preparation and removal of competition by re- BONCINA, A. 2011. History, current status and future prospects of uneven- moving shrub and saw-palmetto (Serenoa repens W. Bartram) sig- aged forest management in the Dinaric region: An overview. Forestry nificantly improves slash pine seedling establishment (Langdon and 84(5):467–478. Bennett 1976) and can be achieved using regular prescribed burning BRADLEY, G.A., AND A.K. KEARNEY. 2007. Public and professional re- (Lohrey and Kossuth 1990, Landers 1991). Though young slash sponses to the visual effect of timber harvesting: Different ways of see- ing. West. J. Appl. For. 22(1):42–54. pine is susceptible to mortality following fire, mature trees are quite BROCKWAY, D.G., AND K.W. OUTCALT. 2010. Comparative analysis of fire-resistant (Brown and Davis 1973). A young cohort of slash pine reproduction techniques for sustainable management of longleaf pine regeneration should not be burned for the first 5 years of age or until forest ecosystems. Report to the Florida Division of Forestry concerning when they are at least 4–5 m high (McCulley 1950, Langdon and the CART Research Project conducted at the Goethe State Forest and Bennett 1976). The seedlings grow fast and in 10–12 years slash Blackwater River State Forest (May 11, Final Report). USDA For. Serv., pine is resistant to fire that does not crown (Wright and Bailey Southern Research Station, Auburn, AL. 55 p. 1982). Additionally, cattle grazing and chemical or mechanical BROCKWAY, D.G., K.W. OUTCALT, J.M. GULDIN, W.D. BOYER, J.L. treatments can be used to reduce the buildup of the fine fuel and WALKER, D.C. RUDOLPH, R.B. RUMMER, J.P. BARNETT,S.JOSE, AND hardwood competition until the new regeneration is resistant to J. NOWAK. 2005. Uneven-aged management of longleaf pine forests: A light fire. scientist and manager dialogue. USDA For. Serv., Gen. Tech. Rep. Slash pine and other species in the southern United States are SRS-78, Asheville, NC. 38 p. difficult to assess using the southern variant of FVS, which lacks a BROWN, A.A., AND K.P. DAVIS. 1973. Forest fire control and use, 2nd ed. full regeneration model. For many other species and forest types McGraw-Hill, New York. 686 p. such as those in eastern Montana; central and northern Idaho; In- CAFFERATA, M.S., AND W.D. KEMPERER. 2000. Economic comparisons be- land Empire; Kootenai, Kaniksu, and Tally Lake; and coastal Alaska tween even-aged and uneven-aged loblolly pine silvicultural systems. Na- tional Council for Air and Stream Improvement, Inc., Tech. Bull. No. and British Columbia, however, the FVS model has full regenera- 801, Research Triangle Park, NC. 88 p. tion establishment extensions that may significantly minimize the CALAMA, R., I. BARBEITO,M.PARDOS,M.DEL RÍO, AND G. MONTERO. uncertainties associated with assumptions regarding natural regen- 2008. Adapting a model for even-aged stands of stone pine (Pinus pinea L.) eration (Dixon 2002, David Lance, USDA Forest Management to complex multi-aged structures. For. Ecol. Manage. 256(6):1390–1399. Service Center, pers. comm., June 26, 2013). CROOKSTON, N.L., AND G.E. DIXON. 2005. The Forest Vegetation Sim- ulator: A review of its structure, content, and applications. Comput. Conclusions Electron. Agric. 49(1):60–80. Overall, our simulation analysis revealed that no single stand DELLA BIANCA, L., AND D.E. BECK. 1985. Selection management in south- conversion scenario led to maximization of all products and services ern Appalachian hardwoods. South. J. Appl. For. 9(3):191–196. simultaneously. There were always tradeoffs involved. The scenarios DIACI, J., G. KERR, AND K. O’HARA. 2011. Twenty-first century forestry: that resulted in higher structural diversity generally led to lower C Integrating ecologically based, uneven-aged silviculture with increased stocks and merchantable timber production. Given the need to demands on forests. Forestry 84(5):463–465. achieve multiple objectives, the scenarios under the assumptions of DICKENS, E.D., J.P. BARNETT, W.G. HUBBARD, AND E.J. JOKELA. 2004. our simulation that would best achieve that mostly require longer Slash pine: Still growing and growing. Proceedings of the slash pine sym- cutting cycles and moderate to high regeneration. posium. USDA For. Serv., Gen. Tech. Rep. SRS-76, Asheville, NC. 145 p. An important observation to note, however, is that the greatest DIXON, G.E. 2002. Essential FVS: A user’s guide to the Forest Vegetation gains for all of the variables in all of the scenarios were made when as Ϫ Simulator. USDA For. Serv., Internal Report, Forest Management Ser- few as 247 seedlings ha 1 were regenerated. Further gains in most of Ϫ1 vice Center, Fort Collins, CO. 244 p. (Revised Dec. 16, 2011.) the scenarios were minor. If as few as 247 seedlings ha can be DONNELLY, D., B. LILLY, AND E. SMITH. 2001. The southern variant of the regenerated in slash pine stands following a harvest, reasonable ben- Forest Vegetation Simulator. USDA For. Serv., Forest Management Ser- efits related to stand structure, C storage, and timber production still vice Center, Fort Collins, CO. 61 p. can be achieved under most of the scenarios. In the worst case EDWARDS, D.M., M. JAY, F.S. JENSEN,B.LUCAS,M.MARZANO,C. scenario of complete failure of regeneration, planting of 247 seed- MONTAGNE´,A.PEACE, AND G. WEISS. 2012. Public preferences across

12 Forest Science • MONTH 2014 Europe for different forest stand types as sites for recreation. Ecology and Forest Vegetation Simulator: Seven things you need to know. USDA For. Society 17(1):27. Serv., Gen. Tech. Rep. NRS-77, Newtown Square, PA. 16 p. FARRAR, R.M. 1996. Fundamentals of uneven-aged management in southern HUANG, C.H., G.D. KRONRAD, AND J.D. MORTON. 2005. The financially pine. Tall Timbers Research Station, Misc. Publ. No. 9, Tallahassee, FL. optimal loblolly pine planting density and management regime for non- 68 p. industrial private forestland in east Texas. South. J. Appl. For. FARRAR, R.M. JR., T.J. STRAKA, AND C.E. BURKHARDT. 1989. A quarter- 29(1):16–21. century of selection management on Mississippi State Farm Forestry Forties. JAMES, F.C., C.A. HESS, AND B.C. KICKLIGHTER. 2001. Ecosystem man- Mississippi State University, Tech. Bull. 164, Mississippi Agricultural agement and the niche gestalt of the red-cockaded woodpecker in lon- and Forestry Experiment Station, Mississippi State, MS. 24 p. gleaf pine forests. Ecol. App. 11:854–870. FLORIDA DIVISION OF FORESTRY. 2007. Ten-year resource management JOHNSON, M.C., D.L. PETERSON, AND C.L. RAYMOND. 2007. Guide to fuel plan for the Tate’s Hell State Forest, Franklin and Liberty counties. treatments in dry forests of the western United States: Assessing forest structure Florida Division of Agriculture and Consumer Services, Division of and fire hazard. USDA For. Serv., Gen. Tech. Rep. PNW-GTR-686, Port- Forestry, Carrabelle, FL. 73 p. land, OR. 322 p. FOOD AND AGRICULTURE ORGANIZATION. 2006. Global forest resources JOHNSON, P.S., S.R. SHIFLEY, AND R. ROGERS. 2009. The ecology and assessment 2010. Food and Agriculture Organization of the United Na- silviculture of oaks, 2nd ed. CABI Publishing, CAB International, Wall- tions, Tech. Bull. 164, Rome, Italy. 340 p. ingford, UK. 580 p. GILMORE, D.W. 2003. To thin or not to thin: Using the Forest Vegetation JOSE, S., E.J. JOKELA, AND D.L. MILLER (EDS.). 2006. Longleaf pine ecosys- Simulator to evaluate thinning of aspen. North. J. Appl. For. 20(1):14–18. tems: Ecology, silviculture, and restoration. Springer-Verlag, New York. GILPIN, T.T., AND J.L. VOWELL. 2006. Restoring hydroperiod on previ- 438 p. ously intensively managed forestlands. P. 94–98 in Proceedings of the KAIPAINEN, T., J. LISKI,A.PUSSINEN, AND T. KARJALAINEN. 2004. Man- international conference, hydrology and management of forested wetlands, aging carbon sinks by changing rotation length in European forests. April 8–12, Tech. Bull. 164, New Bern, NC. Environ. Sci. Policy. 7(3):205–219. GOBSTER, P.H. 1996. Forest aesthetics, biodiversity, and the perceived KELTY, M.J., D.B. KITTREDGE,T.KYKER-SNOWMAN, AND A.D. LEIGHTON. appropriateness of ecosystem management practices. P. 77–97 in Defin- 2003. The conversion of even-aged stands to uneven-aged structure in ing social acceptability in ecosystem management: A workshop proceedings, southern New England. North. J. Appl. For. 20(3):109–116. Brunson, M., L. Kruger, C. Tyler, and S. Schroeder (eds.). USDA For. KERR, G., G. MORGAN,J.BLYTH, AND V. STOKES. 2010. Transformation Serv., Gen. Tech. Rep. PNW-369, Portland, OR. 142 p. from even-aged plantations to an irregular forest: The world’s longest GORTE, R.W. 2009. Carbon sequestration in forests. US Library of Con- running trial area at Glentress, Scotland. Forestry 83(3):329–344. gress, Congressional Research Service Report, Washington, DC. 23 p. KEYSER, C.E. 2008. Southern (SN) variant overview—Forest Vegetation GULDIN, J.M. 2002. Continuous cover forestry in the United States—Ex- Simulator. USDA For. Serv., Internal Rep., Forest Management Service perience with southern pines. P. 296–307 in Continuous cover forestry. Center, Fort Collins, CO. 70 p. (Revised Oct. 4, 2011.) Assessment, analysis, scenarios, Gadow, K., J. Nagel, and J. Saborowski KUULUVAINEN, T., O. TAHVONEN, AND T. AAKALA. 2012. Even-aged and (eds.). Kluwer, Dordrecht, The Netherlands. uneven-aged forest management in boreal Fennoscandia: A review. Am- GULDIN, J.M. 2006. Uneven-aged silviculture of longleaf pine. P. 217–249 bio. 41(7):720–737. in The longleaf pine ecosystem: Ecology, silviculture and restoration, Jose, S., LAIHO, O., E. LAHDE, AND T. PUKKALA. 2011. Uneven- vs. even-aged E.J. Jokela, and D.L. Miller (eds.). Springer, New York. management in Finnish boreal forests. Forestry 84(5):547–556. GULDIN, J.M. 2011. Experience with the selection method in pine stands in LANDERS, J.L. 1991. Disturbance influences on pine traits in the southeast- the southern United States, with implications for future application. ern United States. P 61–95 in Proc. of 17th Tall Timbers fire ecology Forestry 84(5):539–546. conference, May 18–21, 1989. Tall Timbers Research Station, Tallahas- GULDIN, J.M., AND R.M. FARRAR JR. 2002. The plantation conversion see, FL. demonstration at the Crossett Experimental Forest–Implications for LANGDON, O.G., AND F. BENNETT. 1976. Management of natural stands of converting stands from even-aged to uneven-aged structure. USDA For. slash pine. USDA For. Serv., Res. Pap. SE-147, Asheville, NC. 12 p. Serv., Gen. Tech. Rep. SRS-48, Asheville, NC. 6 p. LINDHAGEN, A. 1996. An approach to clarifying public preferences about GUNDERSEN, V.S., AND L.H. FRIVOLD. 2008. Public preferences for forest silvicultural systems: A case study concerning group selection and clear- structures: A review of quantitative surveys from Finland, Norway and cutting. Scand. J. For. Res. 11(1–4):375–387. Sweden. Urban For. Urban Green. 7(4):241–258. LOEWENSTEIN, E.F. 2005. Conversion of uniform broadleaved stands to an HAEFNER, J.W. 2005. Modeling biological systems. Principles and applica- uneven-aged structure. For. Ecol. Manage. 215(1–3):103–112. tions, 2nd ed. Chapman and Hall, New York. 473 p. LOEWENSTEIN, E.F., AND J.M. GULDIN. 2004. Conversion of successionally HAIGHT, R.G. 1987. Evaluating the efficiency of even-aged and uneven- stable even-aged oak stands to an uneven-aged structure. USDA For. Serv., aged stand management. For. Sci. 33(1):116–134. Gen. Tech. Rep. SRS-73, Asheville, NC. 5 p. HAMILTON, D.A. 1991. Implications of random variation in the Stand LOHREY, R.E., AND S.V. KOSSUTH. 1990. Pinus elliottii Engelm. slash pine. Prognosis Model. USDA For. Serv., Res. Note INT-394, Ogden, UT. P. 338–347 in Silvics of North America, volume 1, conifers, agricultural 11 p. handbook 654, Burns, R.M., and B.H. Honkala (tech. coords.). USDA HANEWINKEL, M., AND H. PRETZSCH. 2000. Modelling the conversion For. Serv., Washington, DC. from even-aged to uneven-aged stands of Norway spruce (Picea abies L. MACARTHUR, R.H., AND J.W. MACARTHUR. 1961. On bird species diver- Karst.) with a distance-dependent growth simulator. For. Ecol. Manage. sity. Ecology 42(3):594–598. 134(1):55—70. MAGURRAN, A.E. 1988. Ecological diversity and its measurement. Princeton HENDERSON, J.E. 2008. Uneven-versus even-aged management of south- University Press, Princeton, NJ. 192 p. ern pines: What do landowners give up? What do they gain? For. Land- MARION, W.R., G.W. TANNER, H.K. OBER, AND M. WERNER. 2011. owner July/August:5–8. Management of pine forests for selected wildlife in Florida. Circular 706 HONG, S.H., AND D.J. MLADENOFF. 1999. Spatially explicit and stochastic (Revised), Florida Cooperative Extension Service, Institute of Food and simulation of forest-landscape fire disturbance and succession. Ecology Agricultural Sciences, University of Florida. Available online at 80:81–99. http://edis.ifas.ufl.edu/uw098; last accessed June 10, 2012. HOOVER, C.M., AND S.A. REBAIN. 2011. Forest carbon estimation using the MCCONNELL, W.V. 2002. Initiating uneven-aged management in longleaf

Forest Science • MONTH 2014 13 pine stands: Impacts on red-cockaded woodpecker habitat. Wildl. Soc. 111(3):175–185. Bull. 30(4):1276–1280. SEDJO, R.A. 2001. Forest carbon sequestration: Some issues for forest in- MCCULLEY, R.D. 1950. Management of natural slash pine stands in the vestments. Resources for the Future, Discussion Pap. # 01–34, Wash- Flatwoods of south Florida and north Florida. USDA For. Serv., Cir- ington, DC. 23 p. cular No. 845, Washington, DC. 57 p. SHARMA, A., S. JOSE, K.K. BOHN, AND M.G. ANDREU. 2012. Effects of MCELHINNY, C., P. GIBBONS,C.BRACK, AND J. BAUHUS. 2005. Forest reproduction methods and overstory species composition on under- and woodland stand structural complexity: Its definition and measure- story light availability in longleaf pine–slash pine ecosystems. For. ment. For. Ecol. Manage. 218(1–3):1–24. Ecol. Manage. 284:23–33. MCGAUGHEY, R.J. 1997. Visualizing forest stand dynamics using the stand SHELTON, M.G., AND M.D. CAIN. 2000. Regenerating uneven-aged stands visualization system. P. 248–257 in Proc. of 1997 ASPRS-ACSM-RTI of loblolly and shortleaf pines: The current state of knowledge. For. Ecol. spring convention, Apr. 7–10, Seattle, WA. American Congress on Sur- Manage. 129: 177–193. veying and Mapping, Bethesda, MD. SILVENNOINEN, H., J. ALHO,O.KOLEHMAINEN, AND T. PUKKALA. 2001. MCMINN, J.W. 1981. Site preparation for natural regeneration of slash Prediction models of landscape preferences at the forest stand level. pine. South. J. Appl. For. 5(1):10–12. Landsc. Urban Plan. 56(1–2):11–20. MIZUNAGA, H., T. NAGAIKE,T.YOSHIDA, AND S. VALKONEN. 2010. SMITH, D.M., B.C. LARSON, M.J. KELTY, AND P.M.S. ASHTON. 1997. The Feasibility of silviculture for complex stand structures: Designing stand practice of silviculture: Applied forest ecology. John Wiley and Sons, New structures for sustainability and multiple objectives. J. For. Res. York. 537 p. 15(1):1–2. SMITH, W., P. MILES,C.PERRY, AND S. PUGH. 2009. Forest resources of MURPHY, P.A., AND M.G. SHELTON. 1991. Stand development five years the United States, 2007: A technical document supporting the Forest after cutting to different diameter limits in loblolly-shortleaf stands. P. Service 2010 RPA assessment. USDA For. Serv., Gen. Tech. Rep. 384–393 in Proceedings, 6th biennial southern silvicultural research con- WO-78, Washington, DC. 349 p. ference, Oct. 30–Nov. 1, 1990, Coleman, S.S., and D.G. Neary SORENSEN, D.C., J.A. FINKRAL, E.T. KOLB, AND H.C. HUANG. 2011. (comps.). USDA For. Serv., Gen. Tech. Rep. SE-70, Southeastern For- Short and long term effects of thinning and prescribed fire on carbon est Experiment Station, Asheville, NC. stocks in ponderosa pine stands in northern Arizona. For. Ecol. Manage. NYLAND, R.D. 2002. Silviculture: Concepts and applications, 2nd ed. 261:460–472. McGraw-Hill, New York. 682 p. TAHVONEN, O., T. PUKKALA,O.LAIHO,E.LAHDE, AND S. NIINIMAKI. NYLAND, R.D. 2003. Even- to uneven-aged: The challenges of conversion. 2010. Optimal management of uneven-aged Norway spruce stands. For. For. Ecol. Manage. 172(2–3):291–300. Ecol. Manage. 260(1):106–115. O’HARA, K.L., AND R.F. GERSONDE. 2004. Stocking control concepts in TARP, P., J. BUONGIORNO,F.HELLES, J.B. LARSEN,H.MEILBY, AND N. uneven-aged silviculture. Forestry 77(2):131–143. STRANGE. 2005. Economics of converting an even-aged Fagus sylvatica O’HARA, K.L. 2001. The silviculture of transformation—A commentary. stand to an uneven-aged stand using target diameter harvesting. Scand. For. Ecol. Manage. 151(1–3):81–86. J. For. Res. 20(1):63–74. O’HARA, K.L., AND L.M. NAGEL. 2006. A functional comparison of productivity in even-aged and multi-aged stands: A synthesis for Pinus TECK, R., M. MOEUR, AND B. EAV. 1996. Forecasting ecosystems with the ponderosa. For. Sci. 52(3):290–303. Forest Vegetation Simulator. J. For. 94(12):7–10. HOMAS POMMERENING, A., AND S.T. MURPHY. 2004. A review of the history, T , J.W. 1979. Wildlife habitats in managed forests: The Blue Moun- definitions and methods of continuous cover forestry with special atten- tains of Oregon and Washington. USDA For. Serv., Agri. Handbk. No. tion to afforestation and restocking. Forestry 77(1):27–44. 553, Washington, DC. 512 p. PUKKALA, T., E. LAHDE, AND O. LAIHO. 2010. Optimizing the structure UNITED NATIONS. 2007. Non-legally binding instrument on all types of for- and management of uneven-aged stands in Finland. Forestry 83 ests. United Nations, General Assembly, United Nations Forum on (2):129–142. Forests, New York. 10 p. PUKKALA, T., E. LAHDE,O.LAIHO,K.SALO, AND J. HOTANEN. 2011. A US FISH AND WILDLIFE SERVICE. 2003. Recovery plan for the red- multifunctional comparison of even-aged and uneven-aged forest man- cockaded woodpecker (Picoides borealis), 2nd revision. US Fish and agement in a boreal region. Can. J. For. Res. 41(4):851–862. Wildlife Service, Atlanta, GA. 296 p. REYNOLDS, R.R. 1969. Twenty-nine years of selection timber management on VAN DYCK, M.G., AND E.E. SMITH. 2000. Keywords reference guide for the the Crossett Experimental Forest. USDA For. Serv., Res. Pap. SO-40, Forest Vegetation Simulator. USDA For. Serv., Int. Rep., Forest Man- Southern Forest Experiment Station, New Orleans, LA. 19 p. agement Service Center, Fort Collins, CO. 120 p. (Revised Mar. 2012.) REYNOLDS, R.R., J.B. BAKER, AND T.T. KU. 1984. Four decades of selection VANCLAY, J.K. 1994. Modelling forest growth and yield: Applications to mixed management on the Crossett Farm Forestry Forties. University of Arkansas, tropical forests. CAB International, Wallingford, UK. 312 p. Division of Agriculture, Bull. 872, Agricultural Experiment Station, WEISKITTEL, A.R., D.W. HANN,J.KERSHAW, AND J.K. VANCLAY. 2011. Fayetteville, AR. 43 p. Forest growth and yield modeling. Wiley-Blackwell, Chicester, UK. SAUNDERS, M.R., AND J.E. ARSENEAULT. 2013. Potential yields and 430 p. economic returns of natural disturbance-based silviculture: A case WRIGHT, H.A., AND A.W. BAILEY. 1982. Fire ecology: United States and study from the Acadian Forest Ecosystem Research Program. J. For. southern Canada. John Wiley & Sons, New York. 501 p.

14 Forest Science • MONTH 2014