Western Journal of Applied Forestry

doi: 10.1093/wjaf/28.4.C3pmid: N/A

Open in new tabDownload slide Open in new tabDownload slide Copyright © 2013 by the Society of American Foresters.

Wang,, Mingliang;Stewart, James, D.

doi: 10.5849/wjaf.12-026pmid: N/A

The transition of modulus of elasticity (MOE) values from juvenile to mature wood marks the change between variable, low-quality wood to wood that is stronger and more consistent. Knowing the proportion of mature wood in a log can lower processing costs and allow for higher-quality products. We measured MOE in breast height pith-to-bark samples from lodgepole pine (Pinus contorta) trees in six sites in Alberta and British Columbia, Canada. We assessed eight different two-segment regression models (a first linear, quadratic, exponential or power segment, and a second linear or constant segment) to determine the transition point from juvenile to mature wood based on MOE. All eight models provided useful and significant estimates of the transition point. For the first segment (juvenile phase), the quadratic form predicted the latest transition to mature wood, the exponential form predicted the earliest transition, and the linear and power forms were intermediate. Use of a linear form for the second segment (mature phase) provided only a minor improvement over use of a constant. There were significant differences in transition point based on MOE among some of the sites, and correlations between transition points and tree diameter or height were found at two of the six sites. segmented regression models, wood quality, Pinus contorta Lodgepole pine (Pinus contorta var. latifolia Engelm.) is the most widely distributed conifer in the provinces of British Columbia and Alberta in western Canada. It is also one of the most valuable commercial species harvested in this range, being used in both lumber and pulp production and having wood and fiber attributes suited to production of high-quality and high-value products (Koch 1996). Wood quality varies substantially within the bole, juvenile (pith-associated or crown-formed) wood being more variable and of lower quality for solidwood products than mature (bole-formed) wood, which is stronger and more consistent in nearly all wood attributes (Saranpää 2003). Understanding the transition between juvenile wood and mature wood is an important first step in understanding wood quality. Knowing the point of transition from juvenile to mature wood would permit a processor to optimize the use of a log, milling juvenile wood for low-value commodity lumber (studs) or biomass, and milling mature wood for high-value machine stress-rated (MSR) lumber or veneer. Accurate transition point (TP) prediction from forest inventory information could lead to more efficient use of the log, less production of off-specification lumber and lower costs. Feedstock characteristics can be anticipated and, thus, products identified and segregated early in the supply chain. Mansfield et al. (2007, 2009) estimated transition points (TPs) for ring density, microfibril angle (MFA), and fiber length of fire-origin lodgepole pine trees grown in western Alberta and interior British Columbia. A more recent study compared six different two-segment regression models with respect to their estimated TPs for MFA in a different subset of the lodgepole pine population at six sites in Alberta and British Columbia where long-term silviculture trials were established in natural fire-origin stands (Wang and Stewart 2012). In this study, we focus on transition analysis for another important fiber attribute, modulus of elasticity (MOE), which is related to wood strength and is a key contributor to lumber quality. Juvenile and mature wood are typically demarcated by analyzing trends in wood characteristics that vary with physiological age (defined as the number of rings from the pith), especially ring density (Bendtsen and Senft 1986, Clark et al. 2006, Mansfield et al. 2007, 2009, Tasissa and Burkhart 1998) and fiber length (Bendtsen and Senft 1986, Yang et al. 1986, Yang and Hazenberg 1994, Gartner et al. 1997, Bhat et al. 2001, Mansfield et al. 2009). Although there are various transition determination methods available in the literature (see Goudie and Di Lucca [2004] for a summary), to our knowledge, a clear definition of juvenile versus mature wood has not been established. One statistical method of demarcation that has received much attention is segment-based regression analysis (Bendtsen and Senft 1986, Tasissa and Burkhart 1998, Goudie and Di Lucca 2004, Clark et al. 2006, Mansfield et al. 2007, 2009). This method, also known as “piecewise regression,” partitions the data into different intervals of the independent variable and fits separate regression segments to each interval, with a common join point (TP) at the intersection of adjoining intervals (see Hudson 1966, McGee and Carleton 1970, Gallant and Fuller 1973). However, very few studies have analyzed MOE (e.g., Bendtsen and Senft 1986, Roos et al. 1990, Shepard and Shottafer 1992, Evans et al. 2000), each based on only one of three segmented regression models (LIN-C, LIN-LIN, and Q_C, for details see below Equation 2). Wang and Stewart (2012) found that different segmented models all gave reasonable but differing MFA TP estimates. The objective of this study was to determine the TPs based on MOE using the segment-based regression analysis approach. Specifically, we evaluated four pairs of two-segment models for their ability to delineate the MOE TP between juvenile and mature wood for lodgepole pine. Methods Data Acquisition Core samples were taken at breast height (bh, 1.3 m) from 244 lodgepole pine trees in six long-term silviculture research trials in the central foothills of Alberta and in southeastern British Columbia in 2009. Sample trees were taken from selected treatments and from untreated controls. The treatments and other stand information are presented in Table 1, and more information can be found in Stewart et al. (2006). In each selected plot, 15 trees were selected at random from a subset of trees deemed to be representative of commercially viable trees, i.e., those within two standard deviations of the mean height and diameter. We applied this constraint to ensure that results were not unduly influenced by data outliers, given the limited size of samples analyzed. Table 1. Stand and treatment descriptions for the six study sites in fire-origin stands. Open in new tab Table 1. Stand and treatment descriptions for the six study sites in fire-origin stands. Open in new tab Cores were either taken in the field with an increment corer 12 mm in diameter or sawn from a tree disk. Cores were sealed in plastic bags and kept at 4° C except during processing or transport. In the EvaluTree laboratory of FPInnovations in Vancouver, British Columbia, cores were equilibrated to 8% moisture content before analysis with a SilviScan 3 analyzer. Density, determined by X-ray densitometry, and diffraction, determined from the coefficient of variation of the intensity of the X-ray diffraction profile, were combined to compute the dynamic MOE of each sample (Evans 2006), using the following equation where A and B are statistically determined calibration constants, Icv is the coefficient of variation of the intensity of the X-ray diffraction profile, and D is the wood density from X-ray densitometry. For the purpose of this study, we wished to characterize naturally grown trees, rather than evaluate silviculture treatment effects. Therefore, some trees or observations were excluded from our data analysis because of the significant effects of some treatments within the silviculture trials, specifically, for the two British Columbia sites, Cranbrook (CR) and Parson (PA), observations from 1993 (when the treatments were applied) and later were removed. We note that, for these two sites, trees were more than 65 years of age in 1993, an age well beyond the TP from juvenile to mature wood. This allowed us to be sure of eliminating any treatment effect but still gave a long enough time span of data to be similar to those in the other sites (66–78 years). For Teepee North (TN), Teepee Flat (TF), and MacKay (MK), trees from treatments 1,000 stems/ha (sph) and 4,000 sph (TN site), 1,000 sph (TF site), and 750 sph (MK site) were excluded due to obvious thinning effects on MOE development around (or shortly after) the treatment time, as determined by visual graphical checks. For McCardell (MC), we observed no obvious effects on MOE of either thinning or fertilization, and all trees were retained for analysis. Four additional trees were also removed because of the extreme irregularity of their MOE trends. Ultimately, 195 out of the original 244 trees were used in the data analysis. Segmented Regression Analysis Two-segment regression models have frequently been used to estimate the TP from juvenile to mature wood for wood and fiber attributes, including MOE, in terms of the number of rings from the pith (Bendtsen and Senft 1986, Roos et al. 1990, Evans et al. 2000, Mansfield et al. 2009). The general form of such models can be given as where yi is the fiber attribute measurement of the ith ring from the pith, xi is the ring number, θ1 and θ2 are parameter vectors, x0 is the transition parameter, and ei is the error term. The two lines, represented by y = f1(x) and y = f2(x), meet at x0. For MOE in particular, to the best of our knowledge, only three two-segment models have been used in previous studies: namely, LIN_LIN, consisting of two simple linear [LIN] regression models (Bendtsen and Senft 1986); LIN_C, a first segment consisting of a simple linear regression model and a second consisting of a constant [C] (Roos et al. 1990, Shepard and Shottafer 1992); and Q_C, consisting of a first quadratic [Q] segment and a second constant segment [C] (Evans et al. 2000). It should be noted that, in juvenile wood-mature wood transition analysis using segmented regression, models are primarily used to provide an objective estimate of the TP, not necessarily to achieve a “best” fit to the data. For the purposes of accurately predicting ring-specific wood attributes, models other than segmented ones can be chosen (Jordan et al. 2005). In this study, we considered five additional two-segment models for MOE: EXP_C (EXP denotes an exponential function); EXP_LIN; P_C (P denotes a power function); P_LIN; and Q_LIN. See Table 2 for a complete listing of the models. The EXP_C model was previously used in an analysis of the juvenile-mature TP for MFA (Bhat et al. 2001), and Q_LIN, in an analysis for wood density (Sauter et al. 1999). Similarly, Wang and Stewart (2012) used all the models except P_C and P_LIN to determine TPs for MFA. We are unaware of the use of P_C and P_LIN in previous studies. The choice of the model for the first segment (linear, exponential, power, or quadratic) depends largely on the nature of MOE development at younger physiological ages. Setting the second segment at a constant value (as in the LIN_C, EXP_C, P_C, and Q_C models) is based on the assumption that in mature wood MOE varies little from ring to ring. Conversely, using a simple linear regression model for the second segment assumes a more general MOE developmental pattern in mature wood. Table 2. The eight two-segment regression models considered in this study. Open in new tab Table 2. The eight two-segment regression models considered in this study. Open in new tab Site-Scale Pooled Data Analysis and Individual Tree Analysis Demarcation of the TP can be carried out at the site level (Bendtsen and Senft 1986, Tasissa and Burkhart 1998, Clark et al. 2006) or the tree level (e.g., Mansfield et al. 2007, 2009). We applied all of the eight models (Table 2) at both levels. The least squares fits of all eight models were obtained using SAS Proc NLIN. For pooled data analysis at each site, Student's t-tests of the slope of the second linear segment were used to indirectly determine whether significant differences existed between the model pairs: i.e., LIN_C versus LIN_LIN, EXP_C versus EXP_LIN, P_C versus P_LIN, and Q_C versus Q_LIN. We also visually compared differences in the transition estimates between these paired models and among all the models. The tree-specific estimates of transition ages provide an understanding of within-tree variability and allow further analysis. Specifically, we considered three kinds of analysis. First, paired Student's t-tests (SAS Proc TTEST) were used to compare the difference in transition estimates between each of the four paired models at each site and for all sites combined. Second, analysis of variance (ANOVA) was performed (SAS Proc GLM) to test for differences in transition estimates between sites for each of the eight models. Third, we used correlation and regression analysis (SAS Proc CORR and Proc GLM) to investigate the relationships between tree TPs and tree variables (e.g., dbh and ht). Results Site-Scale Pooled Data Analysis The eight models all provided good fits to the pooled data from each of the six sites (adjusted R2 0.38 to 0.70, Table 3). For each given second segment (whether constant or linear) and each of the six sites, in general, the quadratic model ranked the best, the power the second, the linear the third, and the exponential the fourth in terms of their adjusted R2 values, although the differences were small (less than 0.02 in most cases). The exceptions were for sites TF and MK, for which the power models provided better fits than the quadratic ones, and for site MC, for which the LIN_LIN model provided the best fit. From the results of the Student's t-tests for the slope estimates of the second linear segment, for the TN site, P_LIN may be reduced to P_C and Q_LIN to Q_C, and for the site MK, P_LIN may be reduced to P_C. Otherwise, for each of the four paired models (LIN_C versus LIN_LIN, EXP_C versus EXP_LIN, P_C versus P_LIN, and Q_C versus Q_LIN), the second linear segment significantly improved the model fits relative to the constant segment. Table 3. Model fit evaluation for each of the six study sites. Open in new tab Table 3. Model fit evaluation for each of the six study sites. Open in new tab For the six sites and for each of the three paired models, using a constant for the second segment resulted in later TPs than using a linear segment; the difference ranged from 1 to 13 years depending on the site and model pair (Table 4). The TPs for the paired quadratic models (Q_C versus Q_LIN) for TN and for the paired power models for MK were not significantly different. However, the difference between the paired power models for TN was relatively large (about 5 years). Table 4. Site-scale estimates of TPs for each of the six study sites. Open in new tab Table 4. Site-scale estimates of TPs for each of the six study sites. Open in new tab In general, for each site, and conditional on the second segment, the quadratic segmented models (Q_C and Q_LIN) estimated later TPs than the power segmented models (P_C and P_LIN); both estimated later TPs than the linear segmented models (LIN_C and LIN_LIN); and all estimated later TPs than the exponential segmented models (EXP_C and EXP_LIN). See Figure 1 for an illustrative example (CR site). The only exceptions were for P_C in comparison with Q_C for the two sites TN and TF. The later TPs estimated by the quadratic models were expected since the TPs were calculated as the maximum point of the quadratic curves. Figure 1. Open in new tabDownload slide Observed (solid lines) and modeled (dashed and dotted lines) modulus of elasticity (MOE, in gigapascals) versus ring number, indicating estimated site-level TPs, at site Cranbrook. Explanation of the models is given in the text and in Table 2. Figure 1. Open in new tabDownload slide Observed (solid lines) and modeled (dashed and dotted lines) modulus of elasticity (MOE, in gigapascals) versus ring number, indicating estimated site-level TPs, at site Cranbrook. Explanation of the models is given in the text and in Table 2. Individual Tree Analysis Following Mansfield et al. (2009), we have assured that transition estimates resulting from the eight models are reasonable by graphical visual check. In doing so, some trees were excluded from our tree-level analysis, mainly because their TP estimates were very late, either approaching the age of the tree or beyond it. This is the case in particular for the P_C and Q_C models, in which scatterplots of MOE by ring number for the specific trees in question indicated no sign of leveling off (to the last ring number), and the first segments (power or quadratic) alone fit the data adequately (making it unnecessary to use the second constant segments). Some other extreme transition values (mainly large relative to the others) as indicated by statistical measures (e.g., box-plots) were considered outliers and were excluded as well. Table 5 presents descriptive statistics of the transition estimates obtained from the eight models at the six sites. Table 5. Descriptive statistics of tree-level transition estimates for eight models at each of the six sites. Note: SD = standard deviation, Min = minimum, Max = maximum. Open in new tab Table 5. Descriptive statistics of tree-level transition estimates for eight models at each of the six sites. Note: SD = standard deviation, Min = minimum, Max = maximum. Open in new tab Transition estimates differed between the pairs of models (EXP_C versus EXP_LIN, LIN_C versus LIN_LIN, P_C versus P_LIN, and Q_C versus Q_LIN) and among sites for each given model (Table 5). The trend in the differences between the EXP, LIN, P, and Q models were consistent across sites and remained so at the tree level. Therefore, our tree-level data analysis performed at each site, or pooled across sites, focused on testing for the differences between the paired models and among the sites. Because the TN site had only a small number of trees (only eight trees for seven out of the eight models), this site was excluded from our site-specific tree-level analysis; however, it was kept in the pooled site analysis where appropriate. Comparison of Transition Points Between Model Pairs Paired Student's t-tests were used to compare the difference in transition estimates between paired models for each of the four pairs of models; namely, EXP_C versus EXP_LIN, LIN_C versus LIN_LIN, P_C versus P_LIN, and Q_C versus Q_LIN, by site (excluding TN) and by all sites (pooled data; Table 6). Table 6. Difference in TP estimates between paired models for each of the four pairs of models for tree-level TP estimation, by site and by all sites (pooled data). Open in new tab Table 6. Difference in TP estimates between paired models for each of the four pairs of models for tree-level TP estimation, by site and by all sites (pooled data). Open in new tab Statistically significant differences were found between the pair of EXP models for sites CR, PA, and MC and for all sites (data pooled), and between the pair of LIN models for sites CR and MC and for all sites. Otherwise, there were no significant differences. Site Difference in Tree Transition Points ANOVA was used to test for differences in transition estimates between sites for each of the eight models (Table 7). Overall, significant differences were found among the five sites (excluding TN) for each model (P < 0.01). As an example, Figure 2 presents box-plots of LIN_C estimates for the five sites. Multiple paired comparisons indicated significant differences between CR and each of TF, MK, and MC and between PA and TF. Table 7. ANOVA results of testing for site differences among transition estimates. a Four degrees of freedom for each model. Open in new tab Table 7. ANOVA results of testing for site differences among transition estimates. a Four degrees of freedom for each model. Open in new tab Figure 2. Open in new tabDownload slide Box-plots of tree-level transition estimates (RN = ring number) obtained from model LIN_C for five sites (see Table 1 for site codes). Figure 2. Open in new tabDownload slide Box-plots of tree-level transition estimates (RN = ring number) obtained from model LIN_C for five sites (see Table 1 for site codes). Relationship Between Tree Transition Points and Tree Variables Correlation and stepwise regression analysis were used to investigate the relationship between tree TPs and tree extrinsic variables, including dbh (dbh; 1.3 m), total tree height (ht), slenderness coefficient (sc = ht/dbh), basal area of all trees larger than the subject tree (bal), crown radius, and crown length; all were measured in 2009. Except for sites CR and MC, there was no significant correlation between transition estimates obtained from any of the eight models and the tree extrinsic variables. For the two sites CR and MC, depending on the model, TPs were generally correlated significantly with some of the tree variables: dbh, ht, sc, bal, crown length, and crown radius (not measured at site CR; Table 8). Table 8. Stepwise regression analysis of tree characteristics for predicting TPs. Open in new tab Table 8. Stepwise regression analysis of tree characteristics for predicting TPs. Open in new tab Most commonly, either dbh or ht was the single most significant variable correlated with TPs, except for the EXP models, in which sc and crown length were most significant at site CR, and for EXP_LIN, in which no significant correlation/regression relationship was found at site MC (Table 8). At site CR, correlation was relatively lower (R2 approximately 0.07–0.15), compared with site MC (R2 approximately 0.15–0.29). At site MC, adding a second variable significantly increased R2 values, from approximately 0.28 to 0.34 by adding sc for model LIN_C and from approximately 0.26 to 0.35 by adding bal, for models LIN_LIN. Discussion Model Form All eight (four pairs) two-segment models gave reasonable estimates of the TP from juvenile to mature wood on the basis of MOE. In general, each of the four paired models—(LIN_C, LIN_LIN), (EXP_C, EXP_LIN), (P_C, P_LIN), and (Q_C, Q_LIN)—obtained essentially the same estimates from the viewpoint of practical usage. For progressive (earlier) determination toward the juvenile phase, EXP_C can be used; for conservative (later) estimation toward the mature phase, Q_C may be more appropriate; otherwise, LIN_C and P_C give intermediate results. This is consistent with what was previously found in an examination of TP on the basis of MFA (Wang and Stewart 2012). As with determination based on MFA, the user may choose the model for MOE-based TP, depending on the application of the model (e.g., peeling veneer versus cutting MSR lumber) and the acceptable risk in estimating the attainment of mature wood properties. For example, a certain proportion of lower-quality juvenile wood in a piece of MSR lumber might not affect the grade of the product but would allow greater volume recovery than a more stringent determination. Using a constant instead of a linear function for the second segment for each of the paired models may be justified, based on their fits (Table 3), paired Student's t-tests (Table 6), and graphical comparison for the six sites. Previous studies (Roos et al. 1990, Shepard and Shottafer 1990, Evans et al. 2000, Shepard and Shottafer 1992) also found that using a constant for the second segment was preferable, although they did not compare paired models. Differences Between Site-Level and Tree-Level Models In comparing the site-level TPs estimated in the pooled data analysis (Table 4) to the mean of individual tree estimates for each site (Table 5), we found discrepancies for some models, particularly the P and Q models, at some sites. These discrepancies were mainly due to the differing data analysis we performed; i.e., the pooled data analysis estimated the TP of a single mean MOE-ring number (RN) curve for a given site, whereas the individual tree analysis estimated the mean TP from the individual tree MOE-RN trajectories for the site, after excluding outliers. Nonetheless, the differences in most cases were probably not significant, as the 95% confidence intervals of the pooled site-level estimates and the mean tree-based estimates generally overlapped. Relationship Between MOE Transition Points and MFA Transition Points Previous studies (e.g., Bendtsen and Senft 1986) have shown that transition ages differ by fiber attribute. The same sampled trees used in this study were also used to estimate tree-based MFA TPs, so we were able to directly compare the two sets of TPs. Paired Student's t-tests showed that, except for model EXP_LIN, tree-based TPs for MOE were in general significantly different from those for MFA for each of the other five models EXP_C, LIN_C, LIN_LIN, Q_C, and Q_LIN (we did not consider the two power models in our MFA study). For the five models, the estimated tree TPs based on MOE were generally greater by 1–21 years, on average, than those based on MFA, depending on the model and site. However, since MOE is derived, in part, from MFA in the SilviScan analysis, some correlation between the two transition estimates is to be expected. Correlation analysis indicated that, in general, tree MOE-based TPs were significantly correlated with those based on MFA for each of the six models. The correlation coefficients were approximately 0.26 (P = 0.0509) for site CR and 0.57 (P < 0.0001) for site MC (Figure 3). The uncorrelated variation can be attributed to the variability in density, which is the other variable used to calculate MOE (see Equation 1). Figure 3. Open in new tabDownload slide Scatterplots of TPs based on MOE versus TPs based on MFA obtained from model LIN_C, for sites CR (top) and MC (bottom). Figure 3. Open in new tabDownload slide Scatterplots of TPs based on MOE versus TPs based on MFA obtained from model LIN_C, for sites CR (top) and MC (bottom). Significant correlations between MOE-based TP estimates and tree characteristics were only found for two sites, and the partial R2 values were low, between 0.06 and 0.30, similar to the result for MFA-based TPs in a previous study (0.30; Wang and Stewart 2012). The site MC was the only site where the correlations were significant in both studies, and in the current study was the site with the stronger correlations. The similarity in results between the two studies can be largely attributed to the fact that MOE is calculated in part based on MFA; however, it also indicates that site MC conditions are sufficiently different from the others that the TP is affected by tree characteristics to the greatest extent here. What those site differences are may warrant future study. Radial Distance to Transition Point Although almost all juvenile-mature wood transition modeling has used number of rings from the pith to drive the model, in milling applications, the radial distance from the pith to the TP, rather than the number of rings, is of interest. We calculated the radial distance (the cumulative ring width) from the pith to the TP, and the radial distance for the whole core, for each of the five sites using the LIN_C model (Figure 4). Distances from the pith were around 60 mm for CR and PA, approximately 33 mm for TF and MK, and approximately 45 mm for MC. Using another model would produce different results, as Q and P models predict a greater distance from pith to TP and less mature wood, whereas EXP models predict the reverse. Combining TP predictions with other data, e.g., diameter, can give mill managers an estimate of the proportion of the stand volume that will yield products with particular attributes. Adjusting the sawing pattern can avoid the juvenile wood and optimize the volume of higher-grade products (Jozsa and Middleton 1994). Eventually, this kind of information could allow harvest scheduling to be based, in part, on the size of product that can be milled from the high-value mature wood, as well as on stand age, volume, or piece size. Figure 4. Open in new tabDownload slide Proportions of juvenile (shaded) and mature wood (white) as measured by cumulative ring widths along the radius from pith to bark for five sites. The TP between juvenile and mature wood was estimated by the LIN_C model. Standard error bars are shown, and the mean tree ages at breast height are noted above the bars. Figure 4. Open in new tabDownload slide Proportions of juvenile (shaded) and mature wood (white) as measured by cumulative ring widths along the radius from pith to bark for five sites. The TP between juvenile and mature wood was estimated by the LIN_C model. Standard error bars are shown, and the mean tree ages at breast height are noted above the bars. Mature Wood MOE Values Another application of the TPs estimated by the model is using them to specify a break point for calculating bulk wood quality values. Given the low level of variation in MOE in mature wood, we believe the average MOE of the mature wood is an adequate representation of MOE values after the TP for milling applications. To illustrate this application, we chose model LIN_C to calculate the differences in mature wood MOE between trees and among sites (Figure 5). ANOVA indicated significant differences in MOE in mature wood among the five sites. MSR lumber grades are based, in part, on MOE, with common grade thresholds at 1.5 × 106, 1.6 × 106, 1.8 × 106, and 2.0 × 106 lb/in2 (Anonymous 2009). Adjusting these grade thresholds for the difference in units and for the differences between the static bending test measures on which the grades are based and our SilviScan-derived values (Raymond et al. 2007), the thresholds are approximately 11.4, 12.1, 13.6, and 15.1 GPa. Compared with results from our study, we find that almost all lodgepole pine mature wood meets the lowest of the four MSR grades for MOE, even on the sites with the lowest mean MOE (Figure 5). For three of our sites, 75% or more of the trees would meet the highest MSR grade standard for MOE. Given the relatively high values for the MOE for mature wood, the least conservative model that predicts the earliest TP (EXP_LIN) may be appropriate when the objective is the lower MSR grades or the site average MOE is high (e.g., site CR; Figure 1). Having the ability to model these MOE values allows a mill to anticipate the potential product out-turn from a harvested stand or to select a stand that has the potential to produce products of specific qualities. Figure 5. Open in new tabDownload slide Box-plots of average MOE of mature wood after TPs estimated by model LIN_C for five sites. Figure 5. Open in new tabDownload slide Box-plots of average MOE of mature wood after TPs estimated by model LIN_C for five sites. Conclusion All of the eight two-segment models can be used to objectively estimate TPs based on MOE, and each pair of models gave similar estimates from the viewpoint of practical usage. EXP_C placed the TP closer toward the juvenile phase, Q_C was more conservative with estimates more toward the mature phase; LIN_C and P_C gave intermediate results. The choice of which model would be most suitable for transition demarcation depends on the product requirements and risk assessment of the end user. Overall, there were significant differences in TPs based on MOE among the different sites. For the two sites CR and MC, each having about 60 sampled trees, TPs were in general correlated significantly with some tree variables, in particular, dbh and ht. Future work will look at the effects of silvicultural treatments on transition age and the variation in TP vertically in the stem. " Fieldwork was carried out by Jared Salvail, John Vallentgoed, Jonathan Martin DeMoor, Dominique Lejour, Calvin Strom, Kirsten Mortensen, and Myriam Suard. Comments on an earlier version by Isabelle Duchesne, Richard Krygier, and Peter Newton are greatly appreciated. Discussions with Roger Whitehead, Paul Bicho, Barbara Dalpke, and Maurice Defo helped to shape the direction of this research. Funding for this study was provided by the Canadian Wood Fiber Centre, Canadian Forest Service, Natural Resources Canada. " This article uses metric units; the applicable conversion factors are: meters (m): 1 m = 3.3 ft; millimeters (mm): 1 mm = 0.039 in.; kilograms (kg): 1 kg = 2.2 lb; hectares (ha): 1 ha = 2.47 ac. Literature Cited anonymous . 2009 . SPS 2 Special products standard for machine graded lumber . National Lumber Grades Authority , New Westminster, BC . 1 p. Bendtsen B.A. Senft J. 1986 . Mechanical and anatomical properties in individual growth rings of plantation-grown eastern cottonwood and loblolly pine . Wood Fiber Sci. 18 : 23 – 38 . OpenURL Placeholder Text WorldCat Bhat K.M. Priya P.B. Rugmini P. 2001 . Characterisation of juvenile wood in teak . Wood Sci. Tech. 34 : 517 – 532 . Google Scholar Crossref Search ADS WorldCat Clark A. III Daniels R. Jordan L. 2006 . Juvenile/mature wood transition in loblolly pine as defined by annual ring specific gravity, proportion of latewood, and microfibril angle . Wood Fiber Sci. 38 ( 2 ): 292 – 299 . OpenURL Placeholder Text WorldCat Evans J.W. II Senft J.F. Green D.W. 2000 . Juvenile wood effect in red alder: Analysis of physical and mechanical data to delineate juvenile and mature wood zones . For. Prod. J. 50 ( 7/8 ): 75 – 87 . OpenURL Placeholder Text WorldCat Evans R. 2006 . Wood stiffness by x-ray diffractometry . P. 138 – 146 in Characterisation of the cellulosic cell wall , Stokke D. Groom L. (eds.). Wiley-Blackwell , Ames, IA . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Gallant A.R. Fuller W.A. 1973 . Fitting segmented polynomial regression models whose joint points have to be estimated . J. Amer. Stat. Assoc. 68 : 144 – 147 . Google Scholar Crossref Search ADS WorldCat Gartner B.L. Lei H. Milota M.R. 1997 . Variation in the anatomy and specific gravity of wood within and between trees of red alder (Alnus rubra Bong.) . Wood Fiber Sci. 29 ( 1 ): 10 – 20 . OpenURL Placeholder Text WorldCat Goudie J.W. Di Lucca C.M. 2004 . Modelling the relationship between crown morphology and wood characteristics of coastal western hemlock in British Columbia . P. 308 – 319 in Fourth workshop on the connection between silviculture and wood quality through modeling approaches and simulation software , Nepveu G. (ed.). Equipe de Recherches sur la Qualité des Bois , INRA, Nancy, France . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hudson D.J. 1966 . Fitting segmented curves whose join points have to be estimated . J. Amer. Stat. Assoc. 61 : 1097 – 1125 . Google Scholar Crossref Search ADS WorldCat Jordan L. Daniels R.F. Clark A. III He R. 2005 . Multilevel nonlinear mixed-effects models for the modelling of earlywood and latewood microfibril angle . For. Sci. 51 : 357 – 371 . OpenURL Placeholder Text WorldCat Jozsa L.A. Middleton G.R. 1994 . A discussion of wood quality attributes and their practical implications . Forintek Canada Corp., Special Rep. , Vancouver, BC . 42 p. Koch P. 1996 . Lodgepole pine in North America. Forest Products Society , Madison WI . 1096 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mansfield S.D. Parish R. Di Lucca C.M. Goudie J. Kang K.-Y. Ott P. 2009 . Revisiting the transition between juvenile and mature wood: A comparison of fibre length, microfibril angle and relative wood density in lodgepole pine . Holzforschung 63 ( 4 ): 449 – 456 . Google Scholar Crossref Search ADS WorldCat Mansfield S.D. Parish R. Goudie J.W. Kang K.-Y. Ott P. 2007 . The effects of crown ratio on the transition from juvenile to mature wood production in lodgepole pine in western Canada . Can. J. For. Res. 37 : 1450 – 1459 . Google Scholar Crossref Search ADS WorldCat McGee V.E. Carleton W.T. 1970 . Piecewise regression . J. Amer. Stat. Assoc. 65 : 1109 – 1124 . Google Scholar Crossref Search ADS WorldCat Raymond C.A. Joe B. Evans R. Dickson R.L. 2007 . Relationship between timber grade, static and dynamic modulus of elasticity, and SilviScan properties for Pinus radiata in New South Wales . NZ J. For. Sci. 37 ( 2 ): 186 – 196 . OpenURL Placeholder Text WorldCat Roos K.D. Shottafer J.E. Shepard R.K. 1990 . The relationship between selected mechanical properties and age in quaking aspen . For. Prod. J. 40 ( 7/8 ): 54 – 56 . OpenURL Placeholder Text WorldCat Saranpää P. 2003 . Wood density and growth . P. 87 – 117 in Wood and its biological basis , Barnett J.R. Jeronimidis G. (eds.). Blackwell Oxford , UK . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sauter U.H. Mutz R. Munro B.D. 1999 . Determining juvenile-mature wood transition in Scots pine using latewood density . Wood Fiber Sci. 31 ( 4 ): 416 – 425 . OpenURL Placeholder Text WorldCat Shepard R.K. Shottafer J.E. 1990 . Effect of early release on specific gravity and wood yield of black spruce . For. Prod. J. 40 ( 1 ): 18 – 20 . OpenURL Placeholder Text WorldCat Shepard R.K. Shottafer J.E. 1992 . Specific gravity and mechanical property-age relationships in red pine . For. Prod. J. 42 ( 7/8 ): 60 – 66 . OpenURL Placeholder Text WorldCat Stewart J.D. Jones T.N. Noble R.C. 2006 . Long-term lodgepole Pine silviculture trials in Alberta: History and current results. Can. For. Serv., Nat. Res. Can. and Foothills Model Forest , Edmonton, AB . 184 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tasissa G. Burkhart H. 1998 . Juvenile-mature wood demarcation in loblolly pine trees . Wood Fiber Sci. 30 ( 2 ): 119 – 127 . OpenURL Placeholder Text WorldCat Wang M. Stewart J. 2012 . Determining the transition from juvenile to mature wood microfibril angle in lodgepole pine: A comparison of six different two-segment models . Annals For. Sci. 69 ( 8 ): 927 – 937 . Google Scholar Crossref Search ADS WorldCat Yang K.C. Benson C.A. Wong J.K. 1986 . Distribution of juvenile wood in two stems of Larix laricina . Can. J. For. Res. 16 : 1041 – 1049 . Google Scholar Crossref Search ADS WorldCat Yang K.C. Hazenberg G. 1994 . Impact of spacing on tracheid length, relative density, and growth rate of juvenile wood and mature wood in Picea mariana . Can. J. For. Res. 24 : 996 – 1007 . Google Scholar Crossref Search ADS WorldCat JD Stewart ©2013 Her Majesty the Queen in right of Canada, Canadian Forest Service and The Society of American Foresters.

Prestemon, Jeffrey, P.;Abt, Karen, L.;Potter, Kevin, M.;Koch, Frank, H.

doi: 10.5849/wjaf.12-032pmid: N/A

The mountain pine beetle has killed lodgepole pine and other species of pines in the western United States in an ongoing epidemic. The most heavily affected states are in the interior West: Colorado, Idaho, Montana, and Wyoming, with smaller losses elsewhere. Timber salvage is one response to the epidemic, which could generate revenues for affected landowners and provide wood to forest product manufacturers and, potentially, energy producers. Salvage is occurring, but policymakers have advocated greater rates of such timber removals. To estimate total costs and revenues from salvage and thereby illuminate the economic dimensions of greater salvage removals, we simulated alternative salvage intensity levels on national forests and on other public and private lands where dead standing timber could be potentially recovered and entered into product markets. Data indicate that 19.7 billion cubic feet of standing dead timber are potentially available for salvage, distributed across 20.3 million acres in 12 western states. Simulations on national forests and on lands under other ownerships indicate that positive net revenues (revenues minus costs) could be produced in states with active timber markets on the West Coast and in the northern Rockies, where timber prices would be less depressed by the introduction of large salvage volumes. The central Rocky Mountain states of Colorado, Utah, and Wyoming, which have the largest percentage volume and acreage impacts from salvable standing dead timber, would not generate profitable timber salvage. Simulations of a hypothetical doubling of demand in Colorado and Montana leave Colorado with smaller losses and Montana with larger gains. insects, wood products, markets, epidemic, spatial equilibrium Pine forests of the western United States and Canada have been experiencing high rates of mortality, in part due to an epidemic outbreak of mountain pine beetle (MPB; Dendroctonus ponderosae; Chapman et al. 2012, Meddens et al. 2012). Other contributors to higher rates of timber mortality include wildfires and a variety of less well-known insects and diseases. Few options are available to timberland managers and landowners to prevent mortality, but the visual impact, potentially increased wildfire threat, and the hazard of falling dead trees can be partially addressed by salvaging damaged timber. Indeed, Canadian land managers have embarked on widespread timber salvage as a way of retrieving some value before complete wood decay and sometimes to encourage the regeneration of valuable and productive new forests (Patriquin et al. 2007, Vysea et al. 2009). US national forests are also selling salvaged timber; 14% of all timber sold in fiscal year 2011 was recorded as salvage (USDA Forest Service 2011a). The USDA Forest Service (2011b) has developed the Western Bark Beetle Strategy that lays out priorities and details how increased salvage rates might be achieved. While policymakers and forest managers in the United States have proposed embarking on accelerated rates of salvage on affected lands, such increases require concerted efforts by private landowners to seek out new markets or by public landowners to prioritize management in a way that favors salvage. Complicating the prioritization are weak timber product markets, which could absorb some of the timber removed but that have contracted since the mid-2000s. The number of mills and the total lumber output across the West have declined broadly (Spelter et al. 2009), constraining the potential financial benefits of salvage. Nevertheless, it is plausible that opportunities for positive net revenue or break-even salvage revenue operations exist in certain locations and among particular owner groups in MPB-affected areas of the West. We define net revenues as the revenues received at the mill gate less the costs of harvesting, transportation, and administration. The objective of this study was to identify these opportunities using an economic assessment model that accounts for costs of salvage, the impact of mortality on timber quality, and the effects of salvage on market prices across the West. Our study estimated potential salvage volumes, costs, and revenues resulting from a set of alternative programs that would encourage salvage of standing dead timber; results are summarized by state and by owner groups. Methods In this study, we employ the Economics of Biomass Removals (EBR) model to quantify how the costs and revenues of salvage focused on timberland in the West would be distributed across states and over time. The EBR model was originally designed (Abt and Prestemon 2006, Huggett et al. 2008, Prestemon et al. 2008) to evaluate both the costs and the timber market impacts of alternative programs to reduce hazardous wildfire fuels across the US West and South. Based on this research, the EBR model has been used to evaluate the timber product output implications of hazardous fuel reduction treatments in the West (Barbour et al. 2008), to quantify the jobs and biomass production impacts of these treatments (Abt et al. 2011), and to evaluate whether wildfire hazard reduction treatments yield overall net benefits on timberlands of the West (Prestemon et al. 2012). Model The EBR model is a multiyear two-stage goal and spatial equilibrium program, and it was modified for this study to model the economic feasibility of salvage of dead timber on public and private lands in the West. Although details of the model, including its mathematical formulation, are provided in Prestemon et al. (2008, 2012), describing the modeling framework is important for understanding the current study. EBR can be run for a single year or multiple years, treating timberland (salvaging standing dead trees, in this study) each year according to a predefined set of objectives. After each simulated year, timber inventory data are updated, with the transition to the next year defined by stand growth, new mortality available for salvage, and the volumes removed in the previous year's solution. The first stage of this revised version of the EBR model is a goal program that selects locations in the West to salvage timber by maximizing a goal-weighted sum of salvage volumes, subject to maximum and minimum harvest constraints, a feasible forest product market solution, and an assumed maximum amount of expenditures available to harvest and transport salvage timber to mills. The fundamental unit of information about timber volumes (salvage, nonsalvage) to which the goal weights are applied in the EBR model is the Forest Inventory and Analysis (FIA) plot. To allow for a reasonably fast solution, plot-level information is summed to a spatial and ownership aggregate. Plot-level information includes the average distance to the nearest five sawmills (which consume sawlogs) and the average distance to the two nearest pulp or pole mills (consuming the smaller diameter portions of trees in the stand). Other variables reported or calculated at the plot level include volumes by product category (merchantable live and dead sawlogs and pulpwood) and tree groups—ponderosa pine (Pinus ponderosa and P. lambertiana), lodgepole pine (P. contorta), southern pine (especially P. echinata, P. palustris, P. elliottii, P. taeda), other softwood, and hardwood; ownership (national forest, other public, private); LANDFIRE Map Zone (LANDFIRE 2010); the harvest cost for removing live or dead volumes; and, an administration cost of $200/acre for public and $100/acre for private timberland salvage1. In this study, we aggregated plot-level information up to the map zone for each ownership group for each of the 12 western states in the contiguous United States. LANDFIRE map zones2 are generalized geographical units with similar ecological and biophysical characteristics. The 50 United States contain 79 such zones, which span state boundaries. The 12 states in this study contain 29 map zones, although the area and total standing timber volume found in these zones vary widely. Therefore, the basic modeling units, from which treatment volumes could be obtained in the first stage, are the map zone–ownership aggregates in each of the 12 western states. Depending on the simulation implemented (more information on the simulations is provided in the next section), treatment volumes selected in the first stage could be obtained from parts or all of one map zone–ownership aggregate. Harvest costs, timber volume information by species and live or dead status, and transport costs were expanded to the map zone–ownership aggregate using an area expansion factor. The result is a summary of the total area of stands of salvable timber in each map zone–ownership aggregate and for each of these the weighted average volumes per acre by species by product by live and salvable dead, weighted average transport distances to mills, and the weighted average harvest cost. Finally, the goal weights placed on map zone–ownership aggregates were the presolution net revenues of timber salvage removed; only dead standing salvable timber could be removed. The net revenue in the first stage is defined as the premarket solution value of salvaged sawlogs and pulpwood by species: the delivered volume multiplied by each product's respective market price times a salvage discount factor minus the total stand's harvest and transport costs per acre. As defined, net revenues can be negative. In effect, the EBR model had an ordered preference for salvage of timberlands according to their per acre net revenues. The second stage of the EBR model maximizes, subject to the salvage volumes selected in the first stage in each map zone–ownership aggregate location, the sum of timber product producer and consumer surplus minus transport costs for harvested volumes of both salvage and nonsalvage timber moving across state and international borders (Samuelson 1952, Takayama and Judge 1964). The basic timber product market modeling unit—two levels of aggregation higher than the map zone–ownership—in which equilibrium product prices by species group are obtained, is the state. Trade restrictions that ban exports of roundwood flowing from western US federal lands are imposed. State level maximum processed volumes are determined by state-level mill capacities that act as a physical limit on the volume of timber products that can be processed within the state without new processing capacity being added. We allow these capacities to be exceeded by up to 30% to reflect the possibility of adding shifts to existing mills (Prestemon et al. 2008). The EBR model also allows for the siting of new processing capacity, although this is not implemented endogenously (as in Ince et al. 2008). The second stage optimal solution is a set of market equilibrium product prices, and the volumes by species of timber salvaged (and harvested nonsalvage) produced in each state, consumed at mills in each location, and traded across state and international borders. The result of a set of a multiyear simulations run by the EBR model is an assessment of the net revenue impacts of salvage on national forests, other public, and private lands in the 12 contiguous western US states. In this study, we further summarize the results in terms of salvage costs and salvage revenues by state and ownership group. While not reported in this study, model outputs also include prices and economic welfare changes resulting from changes in salvage. Such changes may be of interest when seeking to quantify how salvage negatively impacts the welfare of owners of nonsalvage timber (e.g., Prestemon and Holmes 2004, Prestemon et al. 2006). It is worth noting, however, that the net revenues generated from salvage on private lands are gross, before taxes. By varying an assumption on the magnitude of a government program to salvage national forest, other public, or private timber, we describe how the geographical focus of a government-subsidized or national forest salvage program might shift across states in the West. By carrying out a “what-if” scenario that tests the effects of a doubling of the total mill capacity in two states of the West that have been heavily affected by the mountain pine beetle—Montana and Colorado—we examine how efforts to encourage or subsidize the consumption of salvaged timber would affect net revenues of salvage obtained by timberland owners (public and private). Finally, by altering our assumption about the regular accumulation of additional standing volumes of salvage timber—from a set annual percentage increase to no more accumulation of standing dead timber—we assess how the spatial targeting of salvage efforts on national forests and other lands would be affected. Data Timber inventory data from Forest Service FIA surveys were assembled for all timberland that is open to harvest and not protected by easements or otherwise set aside for conservation purposes in 12 western US states. The survey years used for each state are reported in Table 1. Data are summarized by owner group (all owners and national forests only), by species group (ponderosa pine, lodgepole pine, other softwood, and hardwood), and by product (sawtimber—representing the cubic foot volume in the sawlog portion of the tree—and pulpwood—representing all other growing stock volume in the tree). Trees coded as standing dead had only total volume measured, so allocations to sawtimber and pulpwood were assumed to be identical to the overall share of sawtimber and pulpwood found in the live trees on the plot, if any. Forested plots without live trees were assumed to have a sawtimber share of standing volume equal to 0.8. Further information on FIA methods can be found in Bechtold and Patterson (2005). Table 1. USDA Forest Service Forest Inventory and Analysis (FIA) survey years, by state. 1 Annualized surveys conducted as specified in Bechtold and Patterson (2005). 2 Periodic survey data used for these states where annualized surveys are incomplete. Open in new tab Table 1. USDA Forest Service Forest Inventory and Analysis (FIA) survey years, by state. 1 Annualized surveys conducted as specified in Bechtold and Patterson (2005). 2 Periodic survey data used for these states where annualized surveys are incomplete. Open in new tab While initial interest was in modeling salvage of only MPB-killed stands, the FIA data did not offer the option to restrict the volume and acreage data for dead timber based on the cause of mortality. While some information on MPB-affected forests in the West is available from aerial detection surveys, the data produced by these surveys (e.g., Backsen and Howell 2013) were not suitable for our study (see Meddens et al. 2012). Although modeling the salvage of timber killed by all causes is not the same as modeling the salvage of MPB-killed trees, salvage operations should be indifferent to the cause of mortality. One advantage of employing FIA data is that plots are measured on a representative sample frame and, therefore, have a level of accuracy that provides greater confidence in simulated salvage programs. FIA plots have representative samples of species, sizes of trees, and site conditions, which allow for accurate assessments of both the materials that can be removed during salvage and the costs of removal of salvable timber. Salvage volume adjustment factors were applied to the standing timber, with an assumption that the net salvable volume was 64% of the volume measured by FIA, which is based on regional averages from Fahey et al. (1986) and Lowell et al. (1992). This adjustment was needed to account for the higher degrade present in standing dead timber. The rate of degrade and volume losses to decay vary across climatic gradients in the western US, but we used a common degrade factor, which provided sufficient detail for this analysis. Lumber milling technology to extract greater volumes from beetle-killed timber is advancing, so the 64% might be considered conservative in the coming years for the United States. Nonetheless, we consider it a reasonable first approximation of the effect of beetle damage on merchantable wood volume and overall value. Timberland classified as salvable in these simulations was required to have at least 300 cubic feet (cf) per acre (ac) of standing dead timber. We labeled this land as salvable timberland, and the volumes on this land as salvable volumes. We note here that available salvage volumes were lower on the private lands (a weighted average across these plots of 459 cf/ac) compared to that available on national forests (664 cf/ac) or other public lands (578 cf/ac) evaluated in this study. Based on previous research (Abt and Prestemon 2006, Huggett et al. 2008, Prestemon et al. 2008), we assumed that timberland with less than 300 cf/ac of standing dead timber would not be profitable to harvest, as only standing dead timber from salvable stands is harvested. Frequently, a portion of the standing inventory in salvable stands is live (green), but our simulations assumed that the live timber was left unharvested. In effect, only stands in which 100% of the standing volume was dead received a final harvest (clearcut), while other stands received partial cuts. This assumption could be relaxed in future modeling but was maintained throughout the simulations reported here. Simulations Simulations were done for 10-year programs of salvage using 20 alternative program sizes, i.e., the overall scale of salvage operations across owner groups and landscapes (Table 2). In this paper, we use the term “program size” to refer to the scale of salvage operations across the owner group and states involved in each simulation. Sizes were measured by the total expenditures made (costs incurred) to harvest and transport the salvaged timber and to administer the salvage sales. Sizes of programs were varied to better understand the importance of market price impacts and the effects of processing capacity constraints in limiting the net revenues achieved from salvage. In reality, program size might refer to specific programs funded by the government to remove salvable timber from public lands or to subsidize such salvage on private lands. To simulate the effect of a continuation of the current MPB epidemic, salvable timberland area was allowed to increase over time in most of our simulation scenarios. This increase was allowed only if the modeling unit (map zone–ownership aggregate in each state) had available timberland with less than 300 cf/ac (about 10.5 tons/ac) of standing dead timber (i.e., “healthy” timberland) from which to draw. If there was timberland from which to draw, then 2% of that timberland was assumed to die and move into the salvable inventory each year3. That is, in modeling units with healthy timberland available, by the end of the simulation run of 10 years (for each simulated program size), 20% of the healthy timberland in that unit was reclassified as dead but salvable in the model. We assumed the epidemic had ended in only two of the simulation scenarios (Table 2). Table 2. Scenarios for this study. 1 “NF” stands for national forests. 2 A continuing epidemic is simulated by moving 2% of healthy timberland into salvable timberland each year. An ending epidemic is simulated by holding area of salvable timberland constant. 3 “2 × pulpwood” means that processing capacity doubles for mills processing small diameter materials (pulpwood). “2 × all” means that processing capacity doubles for mills processing sawtimber and mills processing small diameter materials (pulpwood). Average haul distances to these mills shrink by 30%. Open in new tab Table 2. Scenarios for this study. 1 “NF” stands for national forests. 2 A continuing epidemic is simulated by moving 2% of healthy timberland into salvable timberland each year. An ending epidemic is simulated by holding area of salvable timberland constant. 3 “2 × pulpwood” means that processing capacity doubles for mills processing small diameter materials (pulpwood). “2 × all” means that processing capacity doubles for mills processing sawtimber and mills processing small diameter materials (pulpwood). Average haul distances to these mills shrink by 30%. Open in new tab We also evaluated the effects that changes in mill capacities could have on the resulting salvage programs (Table 2). Such an increase in capacity may not be likely without government intervention, given today's low timber prices and uncertainty about long-run availability of timber salvage. However, this what-if scenario is a way to evaluate the impact of expanded timber product demand in each state that could occur under a plausible scenario in which state or federal governments offer subsidies to facilities that, for example, use wood to create energy. Mill capacities were assumed to double for two states (Colorado and Montana), where either (i) only pulpwood capacity was doubled (to reflect only an increase in capacity for small diameter timber), or (ii) both pulpwood and sawtimber capacity were doubled. We chose these two states because, while both have high proportions of salvable timber, their timber markets are fundamentally different: Montana has a large market for traditional timber products (lumber, posts, poles, pulpwood), while Colorado does not. Capacity doubling was enabled in the simulations by distributing added capacity in new mills across each state in a manner similar to its current distribution of mills. The effect of this doubling was to not only increase the amount of material that could be received at mills in the state but also to reduce by a factor of (1 − 2−0.5), or approximately 30%4, the average haul distance from base level haul distances embedded in the EBR model. Results Evaluation of Standing Dead A summary of volumes and area of timberland with available salvable timber reveals that the central and northern Rocky Mountain states have the largest salvable timberland areas (Figure 1) and the largest total salvable volumes (Figure 2). Salvable timber volume, consistent with recent experience with the MPB epidemic, is highest in Montana, Colorado, Wyoming, and Idaho (Figure 2). Consistent with the assumption of our study, salvable volume shown in Figure 2 consists of standing dead volume on plots with at least 300 cf/ac of dead volume. Volumes are substantially larger on national forests compared to other owner groups and are dominated by lodgepole pine and other softwood species. Figure 1. Open in new tabDownload slide Area of timberland with salvable volumes by owner. Figure 1. Open in new tabDownload slide Area of timberland with salvable volumes by owner. Figure 2. Open in new tabDownload slide Salvable timber volume by state, on national forests and on nonreserved timberland owned by other groups. Figure 2. Open in new tabDownload slide Salvable timber volume by state, on national forests and on nonreserved timberland owned by other groups. As displayed in Figures 1 and 2, the majority of timber and land affected in the 12 western states is on national forests—88% of the total salvable volume and 84% of the total affected area. Across states, the share of the total salvable volume on national forests ranges from 61% in Nevada to 94% in Idaho and in terms of area ranges from 57% in Nevada to 91% in Idaho. The implication of these statistics is that programs that do not include national forest lands would leave large areas in the West unsalvaged. The extent of the damage across states can be seen in Figure 3. Westwide, only 7% of all timber volume is dead and salvable, but in five states (Colorado, Montana, Idaho, Utah, and Wyoming), the salvable volume exceeds 12% of total volume and ranges up to 19%. These five states contain 79% of the total dead salvable volume but only 35% of total live volume in the West. The three West Coast states (California, Oregon, and Washington) have less than 2% of their volume in dead salvable trees. Figure 3. Open in new tabDownload slide Proportion of total timber volume that is salvable by state. Figure 3. Open in new tabDownload slide Proportion of total timber volume that is salvable by state. Of the five states with the highest percentage of mortality losses, four have actual volume losses greater than 2 billion cf (Colorado, Idaho, Montana, and Wyoming). Oregon and Utah have more than 1 billion cf of salvable volume. Of these six states with high salvable volumes, only Idaho, Oregon, and Montana have current timber processing capacity capable of absorbing large quantities of salvage. Simulations by State Results from the state simulations are shown in Table 3. The table reports average volumes per acre, average harvest, transport and administration costs, and average revenues from salvage5,6. As a way of describing whether timber salvage would generate positive or negative net revenues, we calculate the ratio of the delivered price of timber products at the mill gate to the total cost of getting the products to the mill gate—the sum of harvest, transport, and administrative costs. Revenue-cost ratios exceeding 1 indicate that salvage could be done with positive net revenues for the landowner. Table 3. Average salvage volumes, costs, revenues, and revenue-cost ratio for 20 different program sizes, ranging from $10 m/year to $100 m/year, by state and owner with current processing capacity and a continuing epidemic. 1 Costs include harvest, transport, and administrative costs. 2 Revenues are the value of timber products at the mill gate. 3 The revenue to cost ratio is the ratio of the value of delivered timber products to the costs of producing and transporting and administering the timber sale. Open in new tab Table 3. Average salvage volumes, costs, revenues, and revenue-cost ratio for 20 different program sizes, ranging from $10 m/year to $100 m/year, by state and owner with current processing capacity and a continuing epidemic. 1 Costs include harvest, transport, and administrative costs. 2 Revenues are the value of timber products at the mill gate. 3 The revenue to cost ratio is the ratio of the value of delivered timber products to the costs of producing and transporting and administering the timber sale. Open in new tab Table 3 shows the results from the set of scenarios that focus on differences by state for salvage programs on either all ownerships or national forests only (assuming a continuing epidemic and current processing capacity). Volumes of salvage removed on a per acre basis in the simulations ranged from a low of 343 cf/ac in Nevada (about 12 tons/ac) to a high of 1,390 cf/ac (49 tons/ac) in Montana. The highest salvaged volumes per acre were found in the states with highest total and percentage salvage (Colorado, Idaho, Montana, and Wyoming), as well as California and Oregon. Salvage generates positive net revenues in Idaho, Montana, Washington, Oregon, California, and South Dakota (Table 3). These are states with relatively vigorous timber markets relative to the size of the salvage being removed. States where salvage generates revenues that are on average less than salvage costs (includes removal, transport, and administration)—that is, timber stumpage values would be less than or equal to zero—include Colorado and Wyoming, which have large proportions of salvable volume, and Nevada. In both Wyoming and Nevada, prices for delivered salvage timber products are driven so low by salvage programs that they approach zero, also indicating zero stumpage value, while in Utah and Arizona, prices drop nearly to zero. The zero and near zero ratio of delivered product revenue to cost occurs because their costs per acre of salvage are high and their markets are weak. For Wyoming and Colorado, relatively high volumes per acre removed lead to quick saturation of available markets even when total acres harvested is small (Table 3). Simulations of Expansion of Processing Capacity Results from the scenarios evaluating the addition of processing capacity are shown in Table 4. Simulating the expansion of mill capacity to absorb salvage, done only for Colorado and Montana, illustrates several potential effects of enhanced demand on salvage opportunities. Expansions in mill capacities are smaller when only pulpwood consuming mills are considered, but the direction of effect is similar to that shown for the scenario when both sawtimber and pulpwood consuming mills double in their capacity (and increase in the overall landscape density). The smaller benefits of expanding pulpwood capacities are because such timber products have relatively low value relative to their harvest, transport, and administration costs. Policies or programs that are designed to increase the use of wood for energy—whose feasibility we do not directly evaluate here—would likely be directed to increasing the market for biomass, which coincides with the current pulpwood market. The sawtimber plus pulpwood expansion could occur if a policy enhanced profitability at biomass use facilities and if privately financed expansion of sawtimber processing capacity occurred because biomass harvests would remove salvable sawtimber as well. Table 4. Average salvage volumes, costs, revenues, and revenue-cost ratio for 20 different program sizes, ranging from $10 m/year to $100 m/year, for Colorado and Montana by owner under current capacity, doubled pulpwood capacity, and doubled all capacity. 1 Costs include harvest, transport, and administrative costs. 2 Revenues are the value of timber products at the mill gate. 3 The revenue to cost ratio is the ratio of the value of delivered timber products to the costs of producing and transporting and administering the timber sale. Open in new tab Table 4. Average salvage volumes, costs, revenues, and revenue-cost ratio for 20 different program sizes, ranging from $10 m/year to $100 m/year, for Colorado and Montana by owner under current capacity, doubled pulpwood capacity, and doubled all capacity. 1 Costs include harvest, transport, and administrative costs. 2 Revenues are the value of timber products at the mill gate. 3 The revenue to cost ratio is the ratio of the value of delivered timber products to the costs of producing and transporting and administering the timber sale. Open in new tab Timber processing capacity is a constraint in the EBR model, even though timber can be shipped greater distances when capacity is limiting so that increasing capacity is expected to change the outcome of the simulations. Increasing the density of mills across the landscape will reduce transportation costs. The average effect of the density increase was to reduce salvage costs (which include transportation costs) by 1 and 6% for Colorado and Montana, respectively, when all owners are considered, and by 2 and 5% for these states, respectively, when only national forests are modeled (Table 4). Revenues per acre with increased processing capacity are more strongly affected than costs, raising them by 21% for Colorado and 5% for Montana when all owners are considered and by 14% for Colorado when national forests only are considered. Revenues per acre fall slightly for Montana, on the other hand, by about 1% when only national forests are modeled because salvage activity is more heavily concentrated in places where timber prices are driven down somewhat by higher overall supply to the market. The fall in revenues is accompanied by a larger fall in costs per acre, however, increasing the ratio of revenues to cost for national forests in Montana under this scenario. The ratio of revenues to costs increases by about 15% for Colorado and by 29% for Montana when all owners are considered but by only 20 and 7%, respectively, when such programs are conducted only on national forests. When capacity was allowed to double7 for both sawtimber and pulpwood consuming mills in Colorado and Montana, private and other public shares increased slightly, by 2% for Colorado and 3% for Montana (Table 4). Simulations of Continuing and Ending Epidemics The simulation of a Westwide salvage program that did not face a net increase in available salvable timber, consistent with a beetle epidemic that has reached its effective end, is reported in Table 5 and in Figures 4 and 5. Assuming the epidemic ends reduces the revenue-to-cost ratio Westwide for all owners but increases it for national forests. Table 5. Average salvage volumes, costs, revenues and revenue-cost ratio for 20 different program sizes, ranging from $10 m/year to $100 m/year, West-wide by owner under both a continuing epidemic and an ending epidemic. 1 Costs include harvest, transport, and administrative costs. 2 Revenues are the value of timber products at the mill gate. 3 The revenue to cost ratio is the ratio of the value of delivered timber products to the costs of producing and transporting and administering the timber sale. Open in new tab Table 5. Average salvage volumes, costs, revenues and revenue-cost ratio for 20 different program sizes, ranging from $10 m/year to $100 m/year, West-wide by owner under both a continuing epidemic and an ending epidemic. 1 Costs include harvest, transport, and administrative costs. 2 Revenues are the value of timber products at the mill gate. 3 The revenue to cost ratio is the ratio of the value of delivered timber products to the costs of producing and transporting and administering the timber sale. Open in new tab Figure 4. Open in new tabDownload slide Area of timber salvage by state, in a simulation of a Westwide salvage program across program sizes ranging from $50 million/year to $1,000 million/year, all owners, under (A) the base case 2% per year of healthy timberland moving into salvable timberland (epidemic continues), (B) no net increase in healthy timberland moving into salvable timberland (epidemic is over). Figure 4. Open in new tabDownload slide Area of timber salvage by state, in a simulation of a Westwide salvage program across program sizes ranging from $50 million/year to $1,000 million/year, all owners, under (A) the base case 2% per year of healthy timberland moving into salvable timberland (epidemic continues), (B) no net increase in healthy timberland moving into salvable timberland (epidemic is over). Figure 5. Open in new tabDownload slide Area of timber salvage by state in a simulation of a Westwide salvage program across program sizes ranging from $50 million/year to $1,000 million/year, for national forests only, under (A) the base case 2% per year of healthy timberland moving into salvable timberland (epidemic continues), (B) no net increase in healthy timberland moving into salvable timberland (epidemic is over). Figure 5. Open in new tabDownload slide Area of timber salvage by state in a simulation of a Westwide salvage program across program sizes ranging from $50 million/year to $1,000 million/year, for national forests only, under (A) the base case 2% per year of healthy timberland moving into salvable timberland (epidemic continues), (B) no net increase in healthy timberland moving into salvable timberland (epidemic is over). For all timberlands combined, salvage under the largest spending program ($1 billion/year) is about 1.3 million ac per year with an ending epidemic (i.e., no net increase in salvable timberland; Figure 4B) versus more than 1.7 million ac/year under an assumed continuing epidemic (Figure 4A). The main reason for the higher salvage rates for a given level of spending shown in Figure 4B as compared to Figure 4A is because new salvage opportunities (locations with positive net revenues from salvage) emerge in places with higher expected net revenues. Furthermore, the smaller area available to salvage under the no-net-increase scenario diversifies where timber is salvaged. For the largest spending programs, almost all salvage occurs in California, Idaho, Montana, Oregon, and Washington under the base case assumption of a net increase in salvable timber over time. When there is no net increase in salvable timber, salvage in the largest programs additionally moves into Colorado, New Mexico, and Utah. The effect of this diversification is to increase average salvage costs by 31% and to increase average revenues per acre by only 16%, resulting in a drop in the revenue-to-cost ratio shown in Table 5. The case of national forests only (Figure 5) produces a similar spatial pattern, with an even stronger move toward interior West states in larger programs if an ending epidemic is simulated. The effects on salvage from an ending epidemic are also similar, increasing average salvage volumes, costs, and revenues. Average revenues increase by 65% on a per acre basis while average costs increased by only 15%, resulting in a rise in the revenue-to-cost ratio (Table 5), which is opposite of the all-owner case. The larger percentage increase in revenues from this simulation compared to the all-owner simulation is because the all-owner simulation harvests stands with both lower revenue and even lower-cost privately owned stands. In places where large salvable volumes exist, a continuing epidemic that leads to increases in salvage will depress prices. These increases in salvage tend to be in states with weak timber markets in the interior West, which experience significant negative price effects from a large-scale salvage program. When the area and volume of such salvable timber is steady, as could occur when the epidemic is ending, these negative price impacts are dampened, yielding a higher ratio of revenue to costs, partly because lower total salvage volumes enter markets and prices are less depressed. In a program that could occur on all lands, the share of timber volume removed from private and other public lands varies by geography and by scenario (Figure 6). Nevada, with very low volumes of salvable timber, had no private or other public volume removed. For the remaining states, the average percentage removed from private and other public lands averaged, across all simulations, from 4% in Colorado to 57% in Washington. Figure 6. Open in new tabDownload slide Proportion of salvage volume removed by owner group under a simulation allowing for choice between private and other public and national forest managed lands. Figure 6. Open in new tabDownload slide Proportion of salvage volume removed by owner group under a simulation allowing for choice between private and other public and national forest managed lands. Westwide, 37% of the salvage occurred on private and other public land when the epidemic was assumed to continue, but this dropped to only 12% when the epidemic was assumed to have ended. This reflects the fact that timber salvaged from private and other public land is preferred but quickly exhausted compared to national forest salvage. In part this preference is due to our assumption that harvest costs on national forest lands were higher than on private lands. Discussion Epidemic bark beetle populations have resulted in widespread mortality of pines in the western United States, most recently affecting lodgepole pine in the northern Rocky Mountains. The existence of substantial salvable timber in high-mortality locations creates opportunities for revenues on both public and private lands. We used existing FIA plots to quantify salvable timber Westwide. We used the EBR simulation model to quantify the costs of salvage and the potential revenues available on salvage. The model accounted for the negative market effects that significant timber volume salvage would have on prices (e.g., Holmes 1991, Prestemon and Holmes 2000) and the degradation that such timber is assessed on sale. With this information, we described places of potential priority for salvage extraction, based on expected net revenues. Unfortunately, insufficient information exists about the causes of tree death on FIA plots, so the assessment of potential net revenues from salvage focuses on all salvable timber, regardless of whether the salvable trees are in areas of active pine beetle outbreak. We note six implications from our study. First, dead salvable timber lies disproportionately on national forest lands. Five states (Colorado, Wyoming, Montana, Idaho, and Utah) have 75% of all dead salvable timber Westwide, with most of this on national forests. On national forests, the most recent data indicate that salvage volume sold, totaling 352 million bf in fiscal year 2011, represents less than one-fifth of all timber sold, (USDA Forest Service 2011a); our analyses show that much more is still available. Second, salvage in California, Oregon, Washington, Idaho, South Dakota, and Montana could generate positive net revenues, on both private and public lands, across a wide range of potential salvage intensities. The greater the salvage intensity across these states, the larger the net revenues generated. Third, our simulations show that salvage would not generate positive net revenues in the interior western states of Arizona, New Mexico, Utah, Colorado, Nevada, or Wyoming. However, salvaging an acre of timber in Colorado is not as much of a money-losing proposition as it would be in the other interior states. Fourth, it is possible that an expansion of timber processing capacity (resulting in an increase in timber demand) could help alleviate some of the financial losses for owners of dead standing timber in these dry interior western states. Such an expansion could come about as a result of government programs to encourage new demand. Our examination of a doubling of capacity was limited to two states heavily affected by the current MPB epidemic: Colorado and Montana. Assuming doubled capacity in wood consuming mills, net revenues generated by salvage are slightly higher in both states. However, states with less robust markets, such as Colorado and Wyoming, are also faced with the prospect of both costly harvests and immediately saturated timber markets upon harvest. Such increased salvage would, furthermore, negatively impact owners of green timber who are seeking to harvest by forcing down market prices. It is possible that efforts to “grow the pie” by encouraging expanded market opportunities—including traditional forest products and perhaps new bioenergy uses—could mitigate some of the negative market price effects of salvage programs. Such market expansions, however, would have to be very large, resulting in timber prices that are higher than today's prices, to limit losses to landowners conducting the salvage harvests. Further, overall higher salvage rates, regardless of directed government efforts to encourage salvage, face an immediate challenge of weak timber markets, a consequence especially of construction market contraction since the mid-2000s (Howard and Westby 2009, Keegan et al. 2011). Fifth, under a scenario where the epidemic is ending, available salvage is more quickly eliminated in the highest net revenue states, making larger programs move to the loss-generating states of the southern interior West. A continuing epidemic would keep most salvage on the West Coast and in Idaho and Montana. It is not clear whether the current MPB outbreak will continue to expand or whether it is reaching its apex. Scenarios run to evaluate the effects of a MPB control program to suppress a continuing epidemic versus an ending epidemic were run Westwide at $50 million/year program increments, up to $1 billion/year. Regardless of whether salvable timberland increases over time, salvage revenue-to-cost ratios were typically highest in California, Oregon, Washington, Idaho, and Montana. In short, rapidly expanding areas and volumes of salvable timber, particularly in states with robust timber markets (Idaho, Montana), could be salvaged profitably. Much of that salvage could and likely would occur on private lands. Sixth, private and other public lands tend to have more profitably salvaged timberland when compared to national forests, even though private and other public salvable timber accounts for only 21% of current standing salvable volume Westwide. We must reiterate, however, that profits from private salvage are taxed, so not all of such revenues are ultimately captured by private landowners. Smaller efforts to encourage salvage could be successful on these lands. Programs favoring private and other public salvage would likely be most successful at removing salvage in states with significant processing capacity and stronger timber prices, such as California, Oregon, Washington, Idaho, and Montana. On the other hand, it is in these states where the economic incentives for such salvage are apparently already present due to higher overall timber prices. Locations with higher prices are more likely to have stands where salvage revenues can exceed salvage costs. As Holmes (1991) explained, standing dead timber inventory has zero opportunity cost, so as long as it has nonzero stumpage value, it can become part of the supply of marketable timber. Finally, we note that our analyses evaluated only the costs and revenues generated by salvage harvest operations. Other benefits and/or costs of salvage harvesting not considered here—e.g., the potential to alter subsequent wildfire activity (e.g., Fraver et al. 2011, Klutsch et al. 2011, Simard et al. 2011) or enhance jobs and income opportunities (Abt et al. 2011)—could contribute to decisions to undertake salvage programs on any lands, private or public. The USDA Forest Service, private landowners, forestry professionals, and the general public have been concerned for at least the last decade about the extraordinary scale of the mountain pine beetle epidemic and its attendant aesthetic and economic impacts as well as the potential dangers it poses to the public. Our results show that whether the current epidemic proceeds apace or whether it stops, whether new markets emerge for salvaged timber or not, the basic geographical contours of the economic feasibility of salvage are not substantially altered: places where timber product markets are strong are likely to have profitable salvage, while places where product markets are weak would need sizable public expenditures to achieve appreciable reductions in the amount of dead standing timber. Nonetheless, this study provides a first approximation of the net revenues obtainable from salvaging beetle-killed timber across the West in the wake of the current epidemic. Additional advances in our understanding could be made with more accurate information on how the rate of wood deterioration varies across forests of the West, how advances in lumber recovery technology could translate into higher beetle-killed log values at mill gates, and how higher wood prices brought about by a rising construction market or new uses for wood in the energy sector would enable higher net revenues from salvage in the West. Endnotes 1. " As indicated in Abt et al. (2011), the $200/ac fixed cost for salvage on government lands accounts for sale preparation, environmental analysis, and harvest monitoring. On nongovernment timberlands, this cost was set at $100/ac. 2. " Boundaries can be viewed at //landfire.cr.usgs.gov/viewer/. 3. " The 2% rate of expansion in salvable timberland is our estimate of the rate of increase in the area of timberland affected by the ongoing MPB epidemic. This expansion rate is consistent with an estimate that an average of 1.8% of the total area of lodgepole pine forest in the western United States experienced mortality annually between 1997 and 2010 due to MPB. This estimate is based on mortality area estimates from Meddens et al. (2012) and lodgepole pine forest area from Smith et al. (2009). 4. " This emerges from a principle relating nearest neighbor distance to population densities (Blackith 1958). 5. " We do not report a benefit to cost ratio because we do not calculate benefits in terms of economic surplus. 6. " Generally, smaller programs generate the highest net revenues per acre, as timber prices are higher and costs are generally lower. The EBR model selects those locations on the landscape to salvage first based on expected net revenues. 7. " The EBR model does not have an “endogenous” mill capacity feature, such as Tobin's profit-driven Q ratio. Ince et al. (2008), when analyzing the timber product output implications of wildfire hazard reduction treatments in the western United States, described how capacity that is endogenous in this way could encourage expanded demand. To understand the effect of expanded salvage demand that might result from enhanced profits for timber processors in this study, we conducted this sensitivity analysis. " We are grateful to Frederick W. Cubbage, Linda L. Langner, and Kurt Niquidet for their thoughtful reviews of the manuscript and to Frank Sapio and Mark Ambrose for helpful discussions regarding data used in this study. This work was supported in part through Research Joint Venture Agreements 10-JV-11330146-064 and 11-JV-11330146-090 between the USDA Forest Service, Southern Research Station, and North Carolina State University. " This article was written and prepared by a US Government employee on official time, and is therefore in the public domain and not subject to US copyright protection. Literature Cited Abt K.L. Prestemon J.P. 2006 . Timber markets and fire-related fuel treatments in the western US . Nat. Res. Mod. 19 ( 1 ): 15 – 43 . Google Scholar Crossref Search ADS WorldCat Abt K.L. Prestemon J.P. Skog K. Barbour R.J. Hemstrom M. Huggett R.J. Jr. 2011 . Simulations of alternative mechanical fuel treatment programs on western timberland . USDA For. Serv., Res. Pap. RP-SRS-52 . 10 p. OpenURL Placeholder Text WorldCat Backsen J.C. Howell B. 2013 . Comparing aerial detection and photo interpretation for conducting forest health surveys . West. J. Appl. For. 28 ( 1 ): 3 – 8 . Google Scholar Crossref Search ADS WorldCat Barbour R.J. Zhou X. Prestemon J.P. 2008 . Timber product output implications of a national program of mechanical fuel treatments applied on government timberland in the Western States of US . For. Pol. Econ. 10 ( 6 ): 374 – 385 . OpenURL Placeholder Text WorldCat Bechtold W.A. Patterson P.L. 2005 . The enhanced forest inventory and analysis program—National sampling design and estimation procedures . USDA For. Serv., Gen. Tech. Rep. SRS-80 . 85 p. OpenURL Placeholder Text WorldCat Blackith R.E. 1958 . Nearest-neighbor distance measurements for the estimation of animal populations . Ecology 39 ( 1 ): 147 – 150 . Google Scholar Crossref Search ADS WorldCat Chapman T.B. Veblen T.T. Schoennagel T. 2012 . Spatiotemporal patterns of mountain pine beetle activity in the southern Rocky Mountains . Ecology 93 ( 10 ): 2175 – 2185 . Google Scholar Crossref Search ADS PubMed WorldCat Fahey T.D. Snellgrove T.A. Plank M.E. 1986 . Changes in product recovery between live and dead lodgepole pine: A compendium . USDA For. Serv., Res. Pap. PNW-RP-353 . 32 p. OpenURL Placeholder Text WorldCat Fraver S. Jain T. Bradford J.B. D'Amato A.W. Kastendick D. Palik B. Shinneman D. Stanovick J. 2011 . The efficacy of salvage logging in reducing subsequent fire severity in conifer-dominated forests of Minnesota, USA . Ecol. Appl. 21 ( 6 ): 1895 – 1901 . Google Scholar Crossref Search ADS PubMed WorldCat Holmes T.P. 1991 . Price and welfare effects of catastrophic forest damage from southern pine beetle epidemics . For. Sci. 37 ( 2 ): 500 – 516 . OpenURL Placeholder Text WorldCat Howard J.L. Westby R. 2009 . US forest products annual market review and prospects, 2005–2009 . USDA For. Serv., Res. Note FPL-RN-0313 . 9 p. OpenURL Placeholder Text WorldCat Huggett R.J. Jr. Abt K.L. Shepperd W. 2008 . Efficacy of mechanical fuel treatments for reducing wildfire hazard . For. Pol. Econ. 10 ( 6 ): 408 – 414 . Google Scholar Crossref Search ADS WorldCat Ince P.J. Spelter H. Skog K. Kramp A. Dykstra D.P. 2008 . Market impacts of hypothetical fuel treatment thinning programs on federal lands in the western United States . For. Pol. Econ. 10 ( 6 ): 363 – 372 . Google Scholar Crossref Search ADS WorldCat Keegan C.E. Sorenson C.B. Morgan T.A. Daniels J.M. Hayes S.W. Impact of the great recession and housing collapse on the forest products industry in the western United States . For. Prod. J. 61 ( 8 ): 625 – 634 . OpenURL Placeholder Text WorldCat Klutsch J.G. Battaglia M.A. West D.R. Costello S.L. Negrón J.F. 2011 . Evaluating potential fire behavior in lodgepole pine-dominated forests after a mountain pine beetle epidemic in north-central Colorado . West. J. Appl. For. 26 ( 3 ): 101 – 109 . OpenURL Placeholder Text WorldCat Landfire . 2010 . LANDFIRE distribution site . Available online at www.landfire.gov/index.php ; last accessed Oct. 18, 2010. Lowell E.C. Willits S.A. Krahmer R.L. 1992 . Deterioration of fire-killed and fire-damaged timber in the Western United States . USDA For. Serv., Gen. Tech. Rep. PNW-GTR-292 . 27 p. OpenURL Placeholder Text WorldCat Meddens A.J.H. Hicke J.A. Ferguson C.A. 2012 . Spatiotemporal patterns of observed bark beetle-caused tree mortality in British Columbia and the western United States . Ecol. Appl. 22 ( 7 ): 1876 – 1891 . Google Scholar Crossref Search ADS PubMed WorldCat Patriquin M.N. Wellstead A.M. White W.A. 2007 . Beetles, trees, and people: Regional economic impact sensitivity and policy considerations related to the mountain pine beetle infestation in British Columbia, Canada . For. Pol. Econ. 9 ( 8 ): 938 – 946 . Google Scholar Crossref Search ADS WorldCat Prestemon J.P. Abt K.L. Barbour R.J. 2012 . Quantifying the net economic benefits of mechanical wildfire hazard treatments on timberlands of the western United States . For. Pol. Econ. 21 : 44 – 53 . Google Scholar Crossref Search ADS WorldCat Prestemon J.P. Abt K.L. Huggett R.J. Jr. 2008 . Market impacts of a multiyear mechanical fuel treatment program in the US . For. Pol. Econ. 10 ( 6 ): 386 – 399 . Google Scholar Crossref Search ADS WorldCat Prestemon J.P. Holmes T.P. 2000 . Timber price dynamics following a natural catastrophe . Am. J. Agric. Econ. 82 ( 1 ): 145 – 160 . Google Scholar Crossref Search ADS WorldCat Prestemon J.P. Holmes T.P. 2004 . Market dynamics and optimal timber salvage after a natural catastrophe . For. Sci. 50 ( 4 ): 495 – 511 . OpenURL Placeholder Text WorldCat Prestemon J.P. Wear D.N. Holmes T.P. Stewart F. 2006 . Wildfire, timber salvage, and the economics of expediency . For. Pol. Econ. 8 ( 3 ): 312 – 322 . Google Scholar Crossref Search ADS WorldCat Samuelson P. 1952 . Spatial price equilibrium and linear programming . Am. Econ. Rev. 42 ( 3 ): 283 – 303 . OpenURL Placeholder Text WorldCat Simard M. Romme W.H. Griffin J.M. Turner M.G. 2011 . Do mountain pine beetle outbreaks change the probability of active crown fire in lodgepole pine forests? Ecol. Monogr. 81 ( 1 ): 3 – 24 . Google Scholar Crossref Search ADS WorldCat Smith W.B. Miles P.D. Perry C.H. Pugh S.A. 2009 . Forest resources of the United States, 2007 . USDA For. Serv., Gen. Tech. Rep. GTR-WO-78 . 336 p. OpenURL Placeholder Text WorldCat Spelter H. McKeever D. Toth D. 2009 . Profile 2009: Softwood sawmills in the United States and Canada . USDA For. Serv., Res. Pap. RP-FPL-659 . 55 p. OpenURL Placeholder Text WorldCat Takayama T. Judge G.G. 1964 . Equilibrium among spatially separated markets: A reformulation . Econometrica 32 ( 4 ): 510 – 523 . Google Scholar Crossref Search ADS WorldCat USDA Forest Service . 2011a . PTSAR (Administrative Unit Summary) – PTSR202S version FY Awarded w/Offer, Cumulative FY 2011 Q1 to FY 2011 Q4 as of 2011 4th quarter . Available online at www.fs.fed.us/forestmanagement/documents/ptsar/2011/2011_Q1–Q4_PTSR202S.pdf ; last accessed July 31, 2012. USDA Forest Service . 2011b . Western bark beetle strategy: Human safety, recovery and resiliency . Available online at www.fs.fed.us/publications/bark-beetle/bark-beetle-strategy-appendices.pdf ; last accessed Feb. 8, 2013. Vysea A. Ferguson C. Huggard D.J. Roach J. Zimonick B. 2009 . Regeneration beneath lodgepole pine dominated stands attacked or threatened by the mountain pine beetle in the south central Interior, British Columbia . For. Ecol. Manage. 258 ( Suppl. ): S36 – S43 . Google Scholar Crossref Search ADS WorldCat Copyright © 2013 by the Society of American Foresters.

Ball, John, J.;Taecker, Chad, A.

doi: 10.5849/wjaf.11-028pmid: N/A

The mountain pine beetle (Dendroctonus ponderosae) is the most serious pest of ponderosa pine (Pinus ponderosa). Management of this insect involves indirect controls, reducing stand susceptibility, or direct controls, reducing the beetle population. One tactic for reducing bark beetle populations is to fell and treat infested trees in place. Treatments usually involve cutting the tree trunk into sections and covering the infested logs with plastic or coating in diesel oil and rotating them to kill the beetles. These treatments are not always practical due to labor intensity or environmental impact. A study was conducted to determine whether cutting infested trunks into short, 60- to 75-cm lengths, during autumn without further treatment would be sufficient to serve as a population reduction tactic. Groups of infested trees were felled in October 2006 and 2007 with the trunks cut into 60- to 75-cm lengths and left in place. Brood densities were determined in these sections and standing infested trees during the following March and June and emergence in September. Felling infested trees during autumn and cutting into short sections without rotating resulted in 21–24% beetle emergence compared to standing infested trees and may be a practical means of reducing the potential spread of localized beetle populations. mountain pine beetle, Dendroctonus ponderosae, epidemic, sanitation, solar treatments The mountain pine beetle (Dendroctonus ponderosae Hopkins) is a scolytine bark beetle native to western North America from British Columbia, Canada to Sonora, Mexico and this range includes the Black Hills of South Dakota and Wyoming (Wood 1982). It has recently been discovered in western Nebraska (Costello and Schaupp 2011). Mountain pine beetle is considered the most serious pest of ponderosa pine (Pinus ponderosa Dougl. ex Laws). The Black Hills of western South Dakota have almost 480,000 ha of ponderosa pine (Riva et al. 2009) and mountain pine beetle epidemics have been recorded there since the 1890s, each lasting 10 years or longer (Lessard 1986). Ponderosa pine stand morality from mountain pine beetle can be 50% or greater during these epidemics (McCambridge et al. 1982). The current mountain pine beetle epidemic in the Black Hills began in 1997 (Allen and Long 2001). Management of the mountain pine beetle centers on indirect controls, primarily thinning, to reduce the forest stand density and its susceptibility to mountain pine beetle (Fettig et al. 2007). The insect generally colonizes ponderosa pines that are 20–30 cm in diameter or larger measured at 1.4 m above the ground (dbh) (Schmid and Mata 2005), though even trees as small as 7.5 cm dbh may be attacked during epidemics. Ponderosa pine stands in the Black Hills with a basal area greater than 27.5 m2/ha are considered highly susceptible to attack (Schmid and Mata 1992) although the threshold may be even lower (Schmid and Mata 2005). Direct controls are focused on increasing insect mortality (Carroll et al. 2006). These tactics include felling infested trees and processing them at a mill or destroying the tree on site. These population reduction tactics must be completed before the mountain pine beetles emerge. Emergence may begin at the end of June in the Black Hills, though most adults do not emerge until mid July with the typical peak occurring in early August (Schmid 1972). While harvesting and processing the infested trees for wood products may be the most cost-effective approach, it is not always an option. Blue-stain fungi, ophiostomatoid ascomycetes, are closely associated with mountain pine beetles and colonize infested trees (Khadempour et al. 2012). This “blued” wood has little market value in the Black Hills. The forest products industry has set limits on the amount of blue-stain wood it will purchase so there are few opportunities to harvest and process this wood. There are also infested stands in the Black Hills where the trees are too inaccessible for harvest to be economically feasible. Burning infested trees is an option, but cutting and piling logs is time consuming and burning can be restricted due to hazardous conditions (McMullen et al. 1986). Fire bans are common in the Black Hills and piling infested logs with the plan of burning them during the winter or spring is risky as there is the possibility that a ban may prevent the work from being done. When removal or destruction of infested trees is not practical, felling and treating in place becomes an option (Carroll et al. 2006). Few treatment tactics can achieve the control obtained by removal or complete destruction, however, 100% beetle mortality within a tree is not necessary. Hopkins (1905) pointed out in the first documented effort of direct control of mountain pine beetle that it was only necessary to obtain about 75% beetle mortality to prevent infestations from spreading. Felling and peeling the bark from the infested logs was a tactic used by Hopkins (1905) during the 1890s mountain pine beetle epidemic. Peeling the bark from the logs exposes the larvae to environmental stress, loss of food, and predation; however, this procedure is very labor intensive or requires specialized mechanical peelers so peeling is not being used in the Black Hills at this time. Felling and cutting the trunks into sections and treating the logs with insecticides was a procedure used until the 1970s (Stevens and McCambridge 1978) but has limited application under current pesticide regulation. Applying diesel fuel oil on the infested logs is an effective means of reducing larval survival (Mata et al. 2001) but the cost of diesel fuel and environmental concerns limits its application. The wood can also be chipped and either left on site or removed but this procedure is limited to accessible sites and can be expensive. The simplest procedure to managing infested trees is felling and cutting the trunk into logs and exposing them to sunlight (Patterson 1930, Negron et al. 2001). Patterson (1930) found limited success with this method. His efforts involved leaving the entire trunk in one piece with the limbs removed, but the severed trunks had to be rolled frequently to achieve significant mortality. The solar treatments used today usually entails cutting the logs into 1.2–2.4-m sections and either rotating the logs periodically or covering them with plastic sheeting (Leatherman 2001). While these treatments can reduce the mountain pine beetle survival in infested logs, they may not be practical in all situations. It may not be easy to return to remote stands to rotate the logs. Plastic sheeting may also be impractical due to the difficulty of having someone return to the logs for removal and disposal of the plastic as well as the cost of the material if large numbers of infested logs need to be covered. Felling the trees and cutting into logs without further treatment may be the limits of practicality in many circumstances. The optimum period to implement mechanical control tactics is considered to be autumn after the cessation of the flight. Leatherman (2001) noted that infested trees should be felled, regardless of treatment, at the end of the flight period, approximately the beginning of October. Negron et al. (2001) felled infested trees in the spring and then either covered the logs with plastic or rotated them periodically. They suggested that there may be greater efficacy of such treatments if trees are felled in autumn. Custer State Park is located in the southern Black Hills of South Dakota approximately 55 km South of Rapid City, South Dakota. The 960 ha northwestern corner of Custer State Park, known as the Needles, experienced an incipient-epidemic mountain pine beetle population transitioning to an epidemic population by the early 2000s. The Needles is bordered on the north and east by the Black Elk Wildness Area, an area that had experienced nearly 100% mortality of the ponderosa pine from mountain pine beetle during the same time period (Allen and Long 2008). Most of the Needles area of Custer State Park is not readily accessible due to the steep south- and west-facing slopes and rocky terrain. Since many of the infested trees could not be removed from the site and the cost of stacking and covering with plastic was considered too expensive, the decision was made to fell the infested trees during autumn, cut the trunks into logs, and leave in place as a direct control measure to reduce the emerging beetle population the following summer. The logs were cut into short sections, approximately 60–75 cm, and separated by simply pushing each piece away from the adjacent sections. This kept all the sections lying on their side and in contact with the ground. While the desire to scatter the logs for ease of passage through these cut stands was a key reason for cutting into 60– 75-cm lengths rather than the 1.2 m done in other studies, these shorter lengths may also increase mortality. Patterson (1930) indicated he observed greater mortality with solar treatments if the trunks were in contact with the ground rather than having them lay across one another. Steed and Wagner (2004) in a study of bark beetles reproductive success in logs commented that if phloem drying occurs near the ends of the logs, brood production could be decreased by cutting logs into short pieces. The objective of this study was to determine whether felling mountain pine beetle infested trees in autumn and cutting the trunks into 60–75-cm lengths, without further treatment, would significantly reduce the number of adult beetles emerging from the logs the following summer compared to emergence from standing infested trees. Methods Almost all the mountain pine beetle infested trees within the Needles area of Custer State Park were marked and felled during October of 2006 and 2007. Approximately 3,800 infested trees were marked in 2006 and 11,900 in 2007. The infested trees were in groups ranging from 2 trees to more than 250, though groups between 20 and 50 trees were the most common. The trunks of the felled trees were cut into approximately 60–75-cm lengths to a minimum log diameter of 15 cm, the smallest diameter wood generally infested by the beetle (Schmid 1972). The logs were rolled to separate them from adjacent sections and received no further treatments, either peeling or rolling. Five randomly selected groups of infested trees in the Needles area were used for the study in 2006–2007 and another five randomly selected for 2007–2008. The groups were selected each year by assigning a number to each group containing 20–50 infested trees and drawing five numbers. The five study sites for each year were on south to southwestern aspects with a slope between 20 and 40% and between 1,850 and 1,950 m elevation above sea level. The study sites were even-age stands with an average basal area of 25 m2/ha and an average tree diameter of 27 cm dbh. Once the infested trees were felled the stands had less than 20% canopy closure as measured by a sighting tube (Ganey and Block 1994). All the study sites for both years included infested trees that were left standing and infested trees that were felled during October and the trunks cut into 60–75-cm lengths. Samples were collected from the logs of the felled and cut trees and the nearby standing, infested trees during the second week of March, June, and September the year following felling. During each sample date, 20 logs from felled and cut trees and 20 standing infested trees were selected, four from each of the five treated stands. A 15.2 × 30.5 cm section of bark was removed from the top of each log (the sunlit side) and the bottom (the side in contact with the ground). The bark was removed from the middle of each log, about 25–30 cm from the end. A log was sampled only once. The logs that were used for sampling were from a portion of the tree that would have been 1.8 m from the ground, the same height that samples were taken from the standing infested trees. An identical size section of bark was removed from the north and south sides of the standing infested trees during the same time periods. The pieces were removed at a height of approximately 1.8 m. Each standing infested tree was sampled only once. The March and June samples had all live life stages counted while the September samples had only adult emergence recorded. Emergence from the standing trees and logs was not determined from the number of beetles collected from caged logs or trunks but by counting the number of exit holes from a 15.2 × 30.5 cm piece of bark following procedures outlined by Safranyik and Linton (1985). Ventilation holes were excluded and we accounted for more than one beetle possibly emerging from the same exit hole, which frequently occurs (Reid 1963). Data were analyzed using generalized linear model (GLM) and analysis of variance (ANOVA) (SAS/STATR version 9.3 software) comparing means of brood and emergence density between logs and standing trees for all sampling dates (March, June, and September) and both years (2007 and 2008) were separated with Duncan's multiple range test (SAS Institute, Inc. 2011). In addition, the number of live mountain pine beetles recorded from the March and June sampling dates between treated logs and standing trees was compared as well as the aspect of the sample, either the top or bottom of the log or the north or south side of the standing infested tree. An identical comparison was made of adult emergence for the September sampling. A one-way ANOVA was conducted for each sampling date with the four treatment/aspect groups; the north and south samples from standing infested trees and the top and bottom samples from logs. If the F ratio was found to be significant (P < 0.05) then Tukey's honestly significant difference (HSD) test was applied to determine where the differences occurred among means at a sample date. Results and Discussion The mean diameter of the standing infested trees used in this study was 27.1 ± 3.4 cm while the diameter of the logs at midsection was 26.7 ± 4.2 cm. There was no significant difference in mean diameter of the logs and the standing infested trees (t = 0.41, t.05 = 1.96). The mean length of the logs was 68.1 ± 11.2 cm. The mountain pine beetles collected in the March 2007 and 2008 from standing trees and logs were third and fourth instar larvae. During the June 2007 and 2008 the collection was composed of fourth instar larvae, pupae, and adults. The June 2007 sampling contained 22% larvae, 31% pupae, and 47% callow adults. The June 2008 sampling contained a higher percentage of larvae, 41%, and lower percentage of pupae (30%) and callow adults (29%). There was no difference in the percentage composition of life stages between standing infested trees and logs during the June sampling from either year. There was a significant difference between the live brood and emergence densities between the logs and standing trees (Table 1). The infested trees and logs had similar brood densities in the March sampling for both years but the densities of live mountain pine beetles between logs and trees differed in the June sampling (Tables 2 and 3). However, at the June sampling, only the tops of the logs had significantly lower densities of live beetles, while the bottoms of the logs had densities similar to that obtained from the standing infested trees. Table 1. Results of generalized liner model analysis on mountain pine beetle brood or emergence density per 0.05 m2 area of bark from logs or standing trees. Data for all sample dates (March, June, and September) for 2 years (2007 and 2008) were combined in the analysis. Treatments/aspect means with the same letter are not significantly different at the P < 0.05 level. Mean separation tests were conducted using Duncan multiple range test (SAS version 9.3, SAS Institute Inc., Cary NC). Open in new tab Table 1. Results of generalized liner model analysis on mountain pine beetle brood or emergence density per 0.05 m2 area of bark from logs or standing trees. Data for all sample dates (March, June, and September) for 2 years (2007 and 2008) were combined in the analysis. Treatments/aspect means with the same letter are not significantly different at the P < 0.05 level. Mean separation tests were conducted using Duncan multiple range test (SAS version 9.3, SAS Institute Inc., Cary NC). Open in new tab Table 2. Mean (SE) of brood densities (March, June) or adult emergence densities (September) of mountain pine beetles per 0.05 m2 area of bark in 20 infested trees and 20 logs by treatment and aspect for 2007. Means of treatments and aspect or log location for each date followed by the same letter are not significantly different at the P = 0.05 level. Open in new tab Table 2. Mean (SE) of brood densities (March, June) or adult emergence densities (September) of mountain pine beetles per 0.05 m2 area of bark in 20 infested trees and 20 logs by treatment and aspect for 2007. Means of treatments and aspect or log location for each date followed by the same letter are not significantly different at the P = 0.05 level. Open in new tab Table 3. Mean (SE) in brood densities (March, June) or adult emergence densities (September) of mountain pine beetles per 0.05 m2 area of bark in 20 infested trees and 20 logs by treatment and aspect for 2008. Means of treatments and aspects or log location for each date followed by the same letter are not significantly different at the P = 0.05 level. Open in new tab Table 3. Mean (SE) in brood densities (March, June) or adult emergence densities (September) of mountain pine beetles per 0.05 m2 area of bark in 20 infested trees and 20 logs by treatment and aspect for 2008. Means of treatments and aspects or log location for each date followed by the same letter are not significantly different at the P = 0.05 level. Open in new tab The results of the March and June sampling were similar to that obtained in an evaluation of similarly treated mountain pine beetle infested trees in the Black Hills. In a comparison of infested trees felled between February and April and cut into logs (length not specified) and scored longitudinally with a chainsaw, there was a decrease in live brood density in the treated trees relative to that of standing infested trees. However, the difference was not significant in June (Schaupp 2003). A study conducted in Colorado by Negron et al. (2001) on solar treatments of infested ponderosa pine logs cut into 1.2-m lengths concluded that placing logs in one or two layers and covering with plastic resulted in the highest mortality of mountain pine beetle, but the authors suggested if covering with plastic sheeting was impractical, then the logs should be left exposed. Negron et al. (2001) did not find a significant difference in overall survival between rotated logs and logs not rotated, though significantly increased survival was found on the underside of logs not rotated. Negron et al. (2001) contend that little change in brood density occurs after late May. However, Schmid (1972) noted brood densities continued to decline beyond the first of June in standing infested trees. Survivorship curves developed by Knight (1959) indicated that substantial mortality of brood may still occur between April and July. Knight (1967) stated that reasonably good estimates of adult emergence could be made from sampling in April. However, because more accurate predictions could be made from samples taken in July, he used these later samples to develop his sequential sampling procedures for trend predictions. Studies by Schaupp (2003) and Negron et al. (2001) terminated sampling in late June and early July, respectively, before emergence, in an effort to simplify posttreatment brood counts. These investigators suggested that a further reduction in survival may have been observed if treatments had been left in place longer (Negron et al. 2001). In this study we found a significant difference in adult emergence in logs compared to emergence from standing infested trees. The brood density was significantly lower only in the tops of the logs, compared to standing infested trees and the bottoms of logs, in the earlier June sampling. The July to August time period resulted in significantly decreased survival on the underside of the logs. In previous studies using mechanical treatments, varied effects on mountain pine beetle brood were found. For example, Schmid et al. (2001) found 8.3% survival after treating single layer logs with diesel oil. Negron et al. (2001) observed no survival in tops and 75% survival in bottoms of single layer logs that were not rotated. Overall survival in these logs was approximately 25%. However, in their study, the logs were placed so the north aspect of the tree, where brood densities are highest, was oriented to the top and the south aspect oriented to the bottom of the log. The logs were also placed in meadows. In the current study, the logs were not oriented with regard to the aspect of the tree and left in the forest stand from which they were cut, a more practical situation for field operation. The survival in logs was 24% (2007) to 21% (2008) in logs compared to standing infested trees. This strong reduction in survival of mountain pine beetle relative to that in standing trees indicates this treatment can be effective in reducing spread. Conclusion Rotating logs, covering them with plastic, or coating with diesel oil are labor intensive and can have a negative effect on the environment. This study found that simply felling trees and cutting them into logs is a practical and effective way to reduce emergence and lower the potential for spread of localized infestations of mountain pine beetle. " The South Dakota Department of Agriculture (SDDOA) provided funds for this work. We would like to thank Coe Foe and Jessica Halverson, SDDOA, for their field and editorial assistance, and two anonymous reviewers and the associate editor for their constructive comments. " This article uses metric units; the applicable conversion factors are: centimeters (cm): 1 cm = 0.39 in.; meters (m): 1 m = 3.3 ft; square meters (m2): 1 m2 = 10.8 ft2; kilometers (km): 1 km = 0.6 mi; hectares (ha): 1 ha = 2.47 ac. Literature Cited Allen K.K. Long D.F. 2001 . Evaluation of mountain pine beetle activity on the Black Hills National Forest. USDA For. Serv., Biological Evaluation R2-02-22, Rocky Mountain Region, Renewable Resources, Forest Health Management , Lakewood, CO . 9 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Allen K.K. Long D.F. 2008 . Evaluation of mountain pine beetle activity in the Norbeck Area, Black Hills National Forest. USDA For. Serv., Forest Health Evaluation RCSC2-08-03, Rocky Mountain Region, Renewable Resources, Forest Health Management , Lakewood, CO . 8 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Carroll A.L. Shore T.L. Safranyik L. 2006 . Direct control: Theory and practice . P. 155 – 172 in The mountain pine beetle: A synthesis of its biology and management in lodgepole pine , Safranyik L. Wilson B. (eds.). Can. For. Serv., Natural Resources Canada, Pacific Forestry Centre , Victoria, BC, Canada . 304 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Costello S.L. Schaupp W.C. Jr. 2011 . First Nebraska state collection record of the mountain pine beetle, Dendroctonus ponderosae Hopkins (Coleoptera: Curculionidae: Scolytine) . Coleopts. Bull. 65 ( 1 ): 21 – 23 . Google Scholar Crossref Search ADS WorldCat Fettig C.J. Klepzig K.D. Billings R.F. Munson A.S. Nebeker T.E. Negron J.F. Norwark J.T. 2007 . The effectiveness of vegetation management practices for prevention and control of bark beetle outbreaks in coniferous forests of the western and southern United States . For. Ecol. Manage. 238 : 24 – 53 . Google Scholar Crossref Search ADS WorldCat Ganey J.L. Block W.M. 1994 . A comparison of two techniques for measuring canopy closure . West. J. Appl. For. 9 : 21 – 23 . OpenURL Placeholder Text WorldCat Hopkins A.D. 1905 . The Black Hills beetle with further notes on its distribution, life history and methods of control. USDA Bureau of Entomol., Bull. 56 , Washington, DC . 24 p. Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Khadempour L. LeMay V. Jack D. Bohlmann J. Breuil C. 2012 . The relative abundance of mountain pine beetle fungal associates through the beetle life cycle in pine trees . Microb. Ecol. 64 : 909 – 917 . Google Scholar Crossref Search ADS PubMed WorldCat Knight F.B. 1959 . Partial life tables for the Black Hills beetle . J. Econ. Entomol. 52 : 1199 – 1202 . Google Scholar Crossref Search ADS WorldCat Knight F.B. 1967 . Evaluation of forest insect infestations . Annu. Rev. Entomol. 12 : 207 – 228 . Google Scholar Crossref Search ADS WorldCat Leatherman D.A. 2001 . Solar treatment of mountain pine beetle trees. Colorado State Forest Service, Insect Info. Series MPB: No. 1 , Fort Collins, CO . 2 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lessard G. 1986 . Mountain pine beetle mortality in ponderosa pine, Black Hills of South Dakota and Wyoming. USDA For. Serv., Biological Evaluation R2-86-02, Rocky Mountain Region, Timber, Forest Pest and Cooperative Forest Management , Lakewood, CO . 22 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mata S.A. Schmid J.M. Leatherman D.A. 2001 . Diesel fuel oil for increasing mountain pine beetle mortality in felled logs. USDA For. Serv., Res. Note RMRS-RN-13, Rocky Mountain Research Station , Fort Collins, CO . 4 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC McCambridge W.T. Hawksworth F.G. Edminster C.B. Laut J.G. 1982 . Ponderosa pine mortality resulting from a mountain pine beetle outbreak. USDA For. Serv., Res. Pap. RM-235, Rocky Mountain Forest and Range Experiment Station , Fort Collins, CO . 7 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC McMullen L.H. Safranyik L. Linton D.A. 1986 . Suppression of mountain pine beetle infestations in lodgepole pine forests. Can. For. Serv., Pacific Forestry Centre, Info. Rep. BC-X-276 , Victoria, BC, Canada . 20 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Negron J.F. Shepperd W.D. Mata S.A. Popp J.B. Asherin L.A. Schoettle A.W. Schmid J.M. Leatherman D.A. 2001 . Solar treatments for reducing survival of mountain pine beetle in infested ponderosa and lodgepole pine logs. USDA For. Serv., Res. Pap. RMRS-RP-30, Rocky Mountain Research Station , Fort Collins, CO . 11 p. Patterson J.E. 1930 . Control of mountain pine beetle in lodgepole pine by the use of solar heat. USDA For. Serv., Tech. Bull. 195 , Washington, DC . 20 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Reid R.W. 1963 . Biology of the mountain pine beetle, Dendroctonus monticolae Hopkins, in the East Kootenay region of British Columbia. III Interaction between the beetle and its host, with emphasis on brood mortality and survival . Can. Entomol. 93 : 225 – 238 . Google Scholar Crossref Search ADS WorldCat Riva R.J. Moser W.K. Haugan D.D. Josten G.J. Brand G.J. Butler B.J. Crocker S.J. et al. 2009 . South Dakota's forests 2005. USDA For. Serv., Res Bull. NRS-3, Northern Research Station , Newtown Square, PA . 96 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Safranyik L. Linton D.A. 1985 . The relationship between density of emerging Dendroctonus ponderosae (Coleoptera: Scolytidae) and density of exit holes in lodgepole pine . Can. Entomol. 117 : 267 – 275 . Google Scholar Crossref Search ADS WorldCat SAS Institute, Inc . 2011 . SAS/STAT user guide, version 9.3 . SAS Institute, Inc . Cary, NC . Schaupp W.C. 2003 . Mountain pine beetle sanitation at the private/public lands interface, RCSC 7-03 [Letter to Forest Supervisor, Black Hills National Forest], March 19 . USDA For. Serv., Mystic Ranger District Office , Rapid City, SD . 3 p. Schmid J.M. 1972 . Emergence, attack densities, and seasonal trends of mountain pine beetle (Dendroctonus ponderosae) in the Black Hills. USDA For. Serv., Res. Note RM-211, Rocky Mountain Forest and Range Experiment Station , Fort Collins, CO . 7 p. Schmid J.M. Mata S.A. 1992 . Stand density and mountain pine beetle-caused tree mortality in ponderosa pine stands. USDA For. Serv., Res. Note RM-515, Rocky Mountain Forest and Range Experiment Station , Fort Collins, CO . 4 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Schmid J.M. Mata S.A. 2005 . Mountain pine beetle-caused tree mortality in partially cut plots surrounded by unmanaged stands. USDA For. Serv., Res Pap. RMRS-RP-54, Rocky Mountain Research Station , Fort Collins, CO . 11 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Schmid J.M. Mata S.A. Kessler R.R. Popp J.B. 2007 . The influence of partial cutting on mountain pine beetle-caused mortality in Black Hills ponderosa pine stands. USDA For. Serv., Res. Pap. RMRS-RP-68 . 19 p. OpenURL Placeholder Text WorldCat Steed B.E. Wagner M.R. 2004 . Importance of log size on host selection and reproductive success of Ips pini (Coleoptera: Scolytidae) in ponderosa pine slash of northern Arizona and western Montana . J. Econ. Entomol. 97 : 436 – 450 . Google Scholar Crossref Search ADS PubMed WorldCat Stevens R.E. McCambridge W.F. 1978 . Protect your pines…from mountain pine beetle. USDA For. Serv., Misc. Publ., Rocky Mountain Forest and Range Experiment Station , Fort Collins, CO . 20 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Wood S.L. 1982 . The bark and ambrosia beetles of North and Central America (Coleoptera: Scolytidae), a taxonomic monograph . Great Basin naturalist memoirs , No. 6 , Brigham Young University , Provo, UT . 135 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Copyright © 2013 by the Society of American Foresters.

Cerise, Luke, M.;Page-Dumroese, Deborah, S.;McDaniel,, Paul;Mayn,, Cole;Heinse,, Robert

doi: 10.5849/wjaf.12-013pmid: N/A

Site preparation following timber harvests is widely used to increase seedling establishment postharvest. Historically, dozer piling and ripping were the most common forms of site preparation in the Intermountain West. Less commonly, terracing of hill slopes was another form of site preparation on the Bitterroot National Forest in western Montana from 1961–1970 on marginally productive lands. Our objective was to compare soil physical and chemical properties as well as timber productivity as evidenced by diameter-at-breast height (dbh) between terraced and standard-site preparation methods as well as unharvested stands. We collected and analyzed soil samples for bulk density, mineral cations, total C, total N, organic matter, particle size, and pH, forest floor measurements, tree dbh, and ground cover. Even after 45 years, visual soil disturbance in site-prepared stands was still observable with a majority of sites having some degree of compaction or rutting damage. Many soil chemical and physical properties were not significantly different among the two site treatments and the unharvested control stands. However, soil organic matter was significantly lower in the terraced and standard site-prepared stands than in the unharvested stands. Ponderosa pine dbh was greater in the terraced stands than in the nonterraced stands, but understory species diversity was low. The loss of surface soil organic matter and understory species associated with both forms of site preparation is a concern for future forest management. Leaving forest residue during harvest operations, limiting travel routes during management operations, and minimizing forest floor displacement may allow for limited soil impacts on future site-prepared stands. terracing, terracettes, soil management, site productivity, dozer piling Maintaining long-term soil productivity in the Intermountain West is a requirement for sustainable forest management (e.g., Ares et al. 2005). However, heavy equipment used during harvesting and site preparation may cause soil degradation and thereby reduce tree growth. Forest productivity declines due to soil compaction, soil displacement, and brush removal have been noted in the Intermountain West (e.g., Clayton et al. 1987, Ross et al. 1986). In addition, lateral soil displacement and increased bulk density have been associated with decreased pine growth as evidenced by reduced tree diameter-at-breast height (dbh), tree height, and radial growth (Clayton et al. 1987). One major concern with site preparations has been the loss of surface organic matter that can reduce the amount of nitrogen (N) being mineralized into the soil profile (Jurgensen et al. 1997). While limited N on forest sites is a common concern, the loss of organic matter (OM) from periodic stand disturbances, such as thinning, is not well known (Page-Dumroese et al. 2010). Timber harvesting on USDA Forest Service (USDA FS) lands in the Intermountain West was widespread and in some areas very intensive in the 1960s and early 1970s. Most harvest operations during this time used ground-based equipment, such as log skidders and tractors. Many, if not all, of the timber harvests were done by clearcutting the desired timber tree species. On harvesting, seedlings were planted either by hand or machine in areas with gradual slopes (<25%) (S. LaSalle, retired USDA FS Supervisor, pers. comm., Jan. 16, 2012). Site preparation was often conducted to reduce understory competition and make planting easier. At the time, terracing steep mountain slopes was considered an acceptable form of site preparation. Terracing was embraced by the Forest Service, in large part, to meet the agency's mandated policy of 100% seedling establishment and to increase timber production on all managed lands by as much as three times to increase income generated by timber sales (Nie 2008). As a site preparation method, terracing removes vegetation from the planting site thereby reducing seedling competition from understory plant species for water, nutrients, and sunlight. During the 1960s and 1970s roughly 6,500 ha of forested land were terraced on the Bitterroot National Forest (BNF) (Figure 1). These areas were steep (30–65% slope), had marginal soil conditions (low water holding capacity, rocky and thin A-horizon), and limited regeneration success (USDA FS 1970). Other site preparation methods included construction of skid roads, slash removal, and bulldozing or blading of the soil surface to remove understory vegetation. In many cases, terrace site preparation had roughly 40% more soil disturbance compared to these standard mechanical site preparations (S. LaSalle, retired USDA FS Supervisor, pers. comm., Jan. 16, 2012). Figure 1. Open in new tabDownload slide Study sites in the Bitterroot National Forest. Solid polygons indicate different treatment types. Figure 1. Open in new tabDownload slide Study sites in the Bitterroot National Forest. Solid polygons indicate different treatment types. Terracing was thought to increase soil moisture by holding snow longer into the growing season, thereby increasing the effective growing season, and to increase soil water infiltration by decreasing runoff (USDA FS 1970, Querejeta et al. 2000). In addition, terraces were thought to better facilitate mechanical planting and harvesting methods (USDA FS 1970). While mechanical site preparation has been shown to increase the survival rate of conifer seedlings in some cases (McNabb et al. 1993), there are concerns of long-term negative impacts on soil quality, particularly for terraced sites (McNabb et al. 1993, Zlatnik et al. 1999). Initial reports filed in conjunction with National Forest Management Act (National Forest Management Act of 1976 [NFMA]) of 1976 stated that terracing mountain slopes in the Intermountain West would produce long-term soil degradation resulting in large tracts of land devoid of plant life. An earlier comparative study of terraced and nonterraced sites by Zlatnik et al. (1999) found a significant increase in the dbh of planted ponderosa pine (Pinus ponderosa) on terraced locations compared to nonterraced locations. The only significant difference in soil characteristics was an increase in silt content found at terraced locations (Zlatnik et al. 1999). Another study in Utah's Wasatch Mountains by Doty (1970, 1972) on trenched mountain slopes (similar to terracing) to reduce erosion from flooding and overgrazing showed similar results. The authors found that trenching did not significantly increase soil moisture through the summer months but that increased snow accumulation on the trenches did improve vegetation establishment. Similar terrace site preparation was done on the Oregon Cascades and resulted in higher growth and survival of Douglas-fir (Pseudotsuga menzieii) trees (Newton et al. 1974). In addition, the authors also reported that mechanically planted seedlings grew better than hand planted seedlings on their study sites. However, there were also concerns that early gains in tree growth associated with terracing or other forms of mechanical site preparation might not extend through to rotation or thinning age because of changes in soil quality or surface soil displacement (Zlatnik et al. 1999). The longer-term impacts of terracing and mechanical site preparation are relatively unknown on the BNF, particularly as they pertain to ponderosa pine growth and changes in soil quality (bulk density, nutrient content, etc.). The purpose of this study was to evaluate changes in timber productivity as related to changes in understory vegetation and soil physical and chemical properties within terraced plantations in comparison to traditional nonterraced, machine-prepared sites for timber production and unharvested stands. Methods Description of Field Sites This study was conducted on timbered stands in the BNF in southwestern Montana (45.917° N, 114.283° W) shown in Figure 1. We selected sites representing three distinct treatments: (1) clearcut and subsequently terraced sites, (2) standard (dozer piling and ripping) mechanical site preparation (nonterraced), and (3) unharvested stands. Slopes range from 25 to 60%, elevations from 1,220 to 2,000 m, and aspects vary from north to southeast. The climate is semiarid alpine, with hot dry summers and cold winters. Average annual precipitation at the West Fork Bitterroot River station near Conner, Montana, is 480 cm, occurring mostly as spring rains and winter snowpack; the mean annual temperature is 6° C (National Oceanic and Atmospheric Administration [NOAA] 2011). Based on the Soil Survey of the BNF Area and field observations, dominant soil series are Sharrott (coarse-loamy, mixed, superactive, frigid Lithic Haplustepts) and Kellygulch (coarse-loamy, mixed, superactive, frigid Typic Haplustepts) (Soil Survey Staff 2012a, 2012b). These soils are formed in colluvium and slope alluvium derived from metasedimentary rock (Belt Supergroup) and granite, respectively (Soil Survey Staff 2012a). Terraces were constructed with bulldozers creating a terrace bench approximately 2.5 m in width. Bulldozed material was side-cast to create the terrace riser. Ponderosa pine was mechanically planted with a Rocky Mountain single disk trencher that created a furrow 10 cm wide and 46 cm deep along the terrace bench location (S. LaSalle, retired USDA FS Supervisor, pers. comm., Jan. 16, 2012). The terrace risers were not planted. Standard site preparation included dozer piling soil to create microsite conditions for seedlings, and in some cases ripping was done to reduce understory vegetation (primarily pinegrass Calmagrostis rubescens). Experimental Design Twenty-four terraced sites and 24 standard mechanically prepared sites (nonterraced) were selected for soil and vegetation sampling. Ten adjacent unharvested units were also sampled (150+ year old trees with no signs of previous harvest activity). Terraced-bench, terraced-riser, nonterraced, and unharvested sites were randomly selected based on slope and aspect so as not to introduce bias. The number of soil sampling plots at each site varied by stand acreage: four plots on 2–10 ha plantations, five on 11–20 ha plantations, and eight on plantations over 21 ha. As a first assessment, we used the Forest Soil Disturbance Monitoring Protocol (FSDMP) (Page-Dumroese et al. 2009) at each sample plot to assess soil attributes and assign a soil visual disturbance class. The FSDMP determines the proportion of points in each attribute (rutting, compaction, etc.) and the visual disturbance class (0, 1, 2, 3) plus the confidence interval (CI). Primarily designed for preharvest and immediate postharvest sampling, we used FSDMP to assess long-term soil disturbance and vegetation growth responses on these machine prepared sites by pooling 125 sample points from each treatment following Page-Dumroese et al. (2012) to accurately represent each treatment. Forest Floor and Mineral Soil Analyses We collected forest floor and mineral-soil samples by separating the forest floor (organic material above the mineral soil) from a 30 cm diameter plot ring to mineral soil. The separation was relatively distinct with little observed mixing of forest floor materials into mineral soil. Soil samples were then collected using a large diameter (10 cm diameter × 30 cm length) impact coring device described in Page-Dumroese et al. (1999) and split into three depth increments: 0–10 cm, 10–20 cm, and 20–30 cm. On terraced treatments, we sampled an equal number of bench and riser locations. Particle size was determined using the hydrometer method as described in Gee and Bauder (1986) for each soil location (terrace bench, terrace riser, and nonterrace) and depth per harvest unit. Soil samples were dispersed using a combination of physical (electric mixer with stirring paddles) and chemical (sodium-hexametaphosphate) means. No replicates or control samples were analyzed. Remaining soil samples were oven dried at 105° C for 24 hours to remove moisture after air drying and sieved through a 2-mm sieve. Rock fragments and roots were separated and weighed individually. Soil pH was measured on a 2:1 deionized water:soil suspension. No replicate measures were conducted; however, samples outside of ± 10% of within treatment variability were retested. Soil OM was determined by loss-on-ignition after combustion at 375° C for 16 hours following the procedure by Nelson and Sommers (1996). Soil C and N was determined using a Leco Tru-Spec CN analyzer (St. Joseph, MI) on 0.5 g homogenized samples. Homogenization and grinding was achieved using a ball mill. Soil exchangeable Ca, Mg, and K were extracted using 1 M ammonium acetate (Thomas 1982). Ca and Mg were analyzed by atomic absorption spectroscopy using a Perkin Elmer 5100PC Atomic Absorption Spectrometer (Waltham, MA). K was analyzed by flame emission. After filtering the soil for cations, the remaining soil was allowed to dry. A subsample of 2 g was then mixed with 2 M KCl and filtered. The filtered solution was then analyzed for extractable NH4 (Mulvaney 1996). Mineral soil C, N, OM, and cation contents were corrected for coarse-fragment content and extrapolated to a ha basis using the fine-fraction bulk density (e.g., Homan et al. 1995, Kulmatiski et al. 2003). Forest floor samples were oven dried at 60° C until constant mass was attained. In some cases this took up to 72 hours, but most samples attained constant mass within 48 hours. We determined forest floor pH, and split samples for organic cations Ca, Mg, and K, and tested for organic C and N as described above. Forest floor samples were dry-ashed at 450° for 6 hours and leached with 2N HNO3 and assayed by atomic adsorption/flame emission techniques to obtain organic cations (Karam 1993). The organic cation solution was analyzed using a Perkin Elmer 5100PC Atomic Absorption Spectrometer (Waltham, MA). Vegetation Sampling and Analysis Ponderosa pine dbh was measured on 2,145 trees within the three treatment types (terrace, nonterrace, unharvested) using a basal area factor (BAF) of 10 or 15, depending on the amount of ground cover (Green et al. 1992). In an effort to keep the tree ages consistent, trees with a tree ring count of 40–50 years were used for dbh measurements. Tree age was not used for selection of terraced stands because all terracing on the BNF occurred from 1964 to 1971 (Worf 1970) resulting in approximately uniform stand age across terraced sites. Natural regeneration and seed trees were not included in the surveys. To measure ground cover by life form we sampled understory vegetation along 30.5 m long transects at each plot location. In 1 m increments along each transect, ocular estimates of percent cover by life form were made and recorded. Global positioning systems (GPS) coordinates were recorded at each soil sampling location using a Trimble Geo XT 3000 series GPS device and pictures were taken at each point for long-term monitoring of the established plots. Statistical Analysis Analysis of variance (ANOVA) was performed in PC-SAS Version 9.2 (SAS Institute, Inc., Cary, NC). Two-way ANOVA was performed on soil physical and chemical data, along with forest floor data using a general linear model with comparisons between treatment means made using Tukey's multiple range tests. Effects of treatments on measured soil variables were tested for significance at the 0.05 level. All soil data were analyzed by soil location (terrace bench, terrace riser, nonterrace) and soil depth. Each soil depth was compared to each soil location at the same depth, starting with 0–10 cm. Then the soil depths were compared against each other for significance (0–10 cm versus 10–20 cm, etc.). Tree dbh and cover by life form transects were analyzed with one-way ANOVA but were not tested for soil depth. The terraced units were not separated into bench/riser locations because all planted trees are on the terrace bench, and cover by life form transects crossed terrace locations. Tree dbh and cover by life form data were analyzed using treatment type (terrace, nonterrace), soil parent material type, slope, elevation, and aspect. Each variable used for tree dbh and cover by life form was tested for interactions using Tukey's multiple range tests. Results Forest Soil Disturbance Monitoring Visual Assessment Forty-five years after harvesting, the majority of disturbance found on terraced units was due to topsoil displacement and soil compaction caused by the site preparation used to construct the terraced units (Tables 1 and 2). The standard site prepared units exhibited similar disturbance class ratings with topsoil displacement and soil compaction being the two main soil disturbances. Both site treatments had similar forest floor disturbance despite the different site preparation techniques. Unharvested units showed little or no soil disturbance due to lack of harvesting. However, in these units tree tip-ups due to wind partially accounted for the topsoil and forest floor displacement. Table 1. FSDMP soil disturbance class for each treatment type on the BNF. Terraced treatment types have a combination of riser and bench locations within the harvested unit. Values are percent of occurrence for each class. Class 0 is undisturbed conditions and Class 3 represents most disturbed conditions. Open in new tab Table 1. FSDMP soil disturbance class for each treatment type on the BNF. Terraced treatment types have a combination of riser and bench locations within the harvested unit. Values are percent of occurrence for each class. Class 0 is undisturbed conditions and Class 3 represents most disturbed conditions. Open in new tab Table 2. FSDMP visual site assessment measurements for three treatment types on the BNF. Values are percent of occurrence found on each treatment type, and values in parentheses are ± CI. Open in new tab Table 2. FSDMP visual site assessment measurements for three treatment types on the BNF. Values are percent of occurrence found on each treatment type, and values in parentheses are ± CI. Open in new tab Tree Growth Diameter and Soil Cover Comparing ponderosa pine dbh across terraced and nonterraced units showed a significant difference (P < 0.04) between the two site preparation treatments. Trees on terraced units had a dbh that was 2.2 cm larger on average than those of similar age in nonterrace units (Figure 2). The terraced treatment generally had less understory cover than the nonterraced sites (Table 3). Ground cover of shrubs and trees < 2 m tall was significantly less on terraced sites compared to nonterraced sites, whereas other ground cover categories were not significantly different between the two treatment types. Total vegetation cover was very similar between the two site preparation methods (Table 3); unharvested stands had greater understory ground cover than either stand treatment. Species compositions listed in Table 4 were similar for the harvested terraced and nonterraced sites, while unharvested sites showed a slightly different grass and forb composition. Figure 2. Open in new tabDownload slide Ponderosa pine mean dbh values for trees measured on both land management treatments in the BNF. Larger dbh values generally relate to larger total tree biomass estimates and indicate an increase in usable timber (e.g., Ter-Mikaelian et al. 1997). Figure 2. Open in new tabDownload slide Ponderosa pine mean dbh values for trees measured on both land management treatments in the BNF. Larger dbh values generally relate to larger total tree biomass estimates and indicate an increase in usable timber (e.g., Ter-Mikaelian et al. 1997). Table 3. Cover by life form transects as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Values in parentheses are standard error of the mean. Open in new tab Table 3. Cover by life form transects as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Values in parentheses are standard error of the mean. Open in new tab Table 4. Dominant understory vegetation found on each treatment type on the BNF. Ponderosa pine was the main tree species planted. Open in new tab Table 4. Dominant understory vegetation found on each treatment type on the BNF. Ponderosa pine was the main tree species planted. Open in new tab Forest Floor Forest floor biomass, OM content, C, N, Ca, and Mg were similar for the terrace bench and riser sites and the harvested nonterraced stands (Table 5). However, in the nonterraced stands, K was significantly higher than in the terrace locations but was equal to that found in the unharvested stands. The unharvested stands had greater forest floor biomass, OM, C, Ca, and Mg than the treated stands. Interestingly, forest floor N was unaffected by treatment type. Table 5. Forest floor measurements as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Values in parentheses are standard error of the mean. Open in new tab Table 5. Forest floor measurements as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Values in parentheses are standard error of the mean. Open in new tab Soil Bulk Density Table 6 shows significant total bulk density differences among sample locations (bench, riser, no terrace) and soil depth, the main differences being at the 0–10 cm and the 20–30 cm depth. Nonterrace (1.35 Mg/ha) and terrace riser (1.29 Mg/ha) locations had similar total bulk densities, but the terrace bench (1.53 Mg/ha) location showed an increase in total bulk density at the 0–10 cm depth. However, total bulk density decreased with depth on the terrace bench location, but increased on the other two treatment types (Table 6). Fine fraction bulk density indicated no significant differences between the different site treatments. The only soil depth that was found to be significantly different was the 10–20 cm depth among the site treatments where harvested nonterrace had a lower fine-fraction bulk density than the other site treatments. Table 6. Soil total bulk density, fine fraction bulk density, and particle size distribution as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Particle size distribution represents entire 30 cm soil core. Values in parentheses are standard error of the mean. Open in new tab Table 6. Soil total bulk density, fine fraction bulk density, and particle size distribution as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Particle size distribution represents entire 30 cm soil core. Values in parentheses are standard error of the mean. Open in new tab Soil OM, Carbon, Nitrogen, and pH Soil OM was significantly higher in unharvested soils at all soil depths compared to site prepared locations (Table 7). At the 0–10 cm depth OM content was 19.2% in the surface soil of the unharvested stands and 7.6% in the nonterraced soil as compared to 4.7% and 5.6% in the soil of the terrace bench and riser locations, respectively. At the 10–20 cm depth there was no significant difference in OM for each of the harvested sample locations (bench, riser, nonterrace). The OM at the 20–30 cm soil depth was not significantly different between the terrace bench location and unharvested stands, but they were both significantly different from the terrace riser and nonterraced sample locations. Table 7. Soil organic matter and pH by depth and land management treatment type on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types at each depth. Values in parentheses are standard error of the mean. Open in new tab Table 7. Soil organic matter and pH by depth and land management treatment type on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types at each depth. Values in parentheses are standard error of the mean. Open in new tab There was a significant decline in total C with increasing soil depth in all treatment types. However, no significant differences among the treatment types were found (Table 8). Soil C was higher in the riser and bench locations but was not significantly different from the nonterraced treatment. Soil N was significantly different between sample location and depth. The terrace riser and bench locations were significantly higher in soil N compared to the nonterraced locations at the 0–10 cm and 10–20 cm depths. The 20–30 cm depth was significantly different across all treatment types and sample locations. No significant differences were found in soil pH across the treatments (Table 7). Table 8. Soil nutrients as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Values in parentheses are standard error of the mean. Open in new tab Table 8. Soil nutrients as affected by land management treatment on the BNF. Within each column, means followed by the same letter are not significantly different (P < 0.05) from different treatment types. Values in parentheses are standard error of the mean. Open in new tab Discussion Forty-five years after site preparation we were still able to detect changes in soil visual classes on both terraced and harvested nonterraced treatments. Most of the soil disturbance on these sites was topsoil displacement and surface compaction (Tables 1 and 2). Topsoil displacement and compaction were found to be light or moderate based on the FSDMP results. Harvested and site prepared stands had a majority of Visual Disturbance Classes 1 and 2; whereas the soil in the unharvested stand was primarily in Disturbance Class 0. Given that terraced sites were 100% disturbed initially, these results suggest that some soil recovery has taken place since. Similarly, standard site prepared stands had approximately 60% soil disturbance immediately after harvesting. Visual classification systems define the attributes and severity class of the disturbance, but ultimately they must reflect the impact of site-specific vegetation growth and hydrologic function (Page-Dumroese et al. 2012). Although we could still detect visual disturbance attributes (Table 1) in the harvested stands after 45 years, there was little or no difference in the measured soil properties. In addition, dbh (Figure 2) did not seem to be diminished due to the soil disturbances, and understory vegetation was present on all of our sites, although not as diverse as the unharvested stands (Table 4). Despite the intensive nature of terracing, tree dbh is actually higher on terraced units in the BNF compared to traditional mechanical site prepped units of the same age class, which is consistent with the findings by Zlatnik et al. (1999). Greater dbh in the terraced stands may be partially attributed to mechanical planting versus hand planting on the nonterraced locations considering that soil conditions were comparable across harvested treatments. Soil conditions were only minimally affected by both mechanical treatments and on these soil types (rocky with a thin A-horizon) do not seem to reduce soil nutrients (Table 8). Bulk density differences between the machine prepared units were not significant and did not seem to limit tree growth. Other studies have also found that terraces tend to have similar bulk densities as their nonterraced counterparts (Querejeta et al. 2000, Ternan et al. 1996). However, we did observe that the terrace bench location had a 12% decrease in bulk density from 0–10 cm depth to 20–30 cm depth compared to a 16% increase in bulk density with depth on the terrace riser, 8% increase in bulk density with depth on the nonterraced units, and a 28% increase in bulk density with depth on the unharvested units (Table 6). Decreased bulk density along the terrace bench location may be partially attributed to subsoiling actions during the terrace construction. We found that soil OM content was generally greater at lower depths on the terrace bench compared to the same depths in terrace riser and nonterrace locations suggesting that topsoil was redistributed across the hill slope and that the initial organic horizons were mixed into the soil profile. Above all, the soil types present on the BNF are very rocky and coarse textured and may not be as susceptible to compaction as finer textured soils (e.g., Andisols). Overall, soil nutrients were similar across the mechanical site preparation treatments after 45 years. Soil nutrients were similar at each depth when compared among the mechanically prepared sites (Table 8); although C was slightly higher in the mineral soil of the unharvested stands than in the mechanically treated units. This indicates that these soils were relatively nutrient limited from the beginning, and therefore, harvesting impacts were minimal. Similar soil nutrient results were found on a site-specific study done in the West Fork Ranger District of the BNF by Zlatnik et al. (1999). The authors also mentioned that if sufficient nutrients had been released from the decomposition of OM on terraced and harvested nonterraced units, there would be no observable differences in soil nutrients. This appears to hold true for our study on the BNF and is likely due to similar forest floor biomass and nutrient concentrations on both the terraced and harvested nonterraced units (Table 5). Given these results, it appears that soil nutrient additions through needle fall and OM decomposition will be comparable among site prepared treatments provided that there is stand regeneration. Long-term impacts of mechanical site preparation can be seen in the reduction of forest floor biomass (Table 5) where even after 45 years of site recovery, the forest floor biomass in the harvested stands was 40% less than the unharvested stands. More importantly, species composition and diversity differed with site treatment. For example, Idaho fescue, junegrass, sulfur buckwheat, and prince's plume are only found in the unharvested stands (Table 4). Idaho fescue was previously found on nonterraced sites by Zlatnik et al. (1999) but was not found in the current assessment. In addition, the authors found ninebark on terraced sites, but none was observed on our terraced sites. Spotted knapweed was very prevalent on our harvested sites (terraced and nonterraced) but was not observed in the unharvested mature stands, nor was it noted in the study by Zlatnik et al. (1999). Spotted knapweed is an invasive species that invades disturbed areas and degrades desirable plant communities (e.g., Watson and Renny 1974). While differences in species composition and diversity may be partially explained by different successional stages in understory vegetation between the site-prepared treatments and the control, we do find it interesting to note that site preparation may encourage knapweed invasion given that the disturbance occurred 45 years ago (though we do not have enough replicates to statistically infer this). Considering the long-term effects of invasive species, the presence of spotted knapweed begs the question whether increased erosion and degraded soil properties may negatively affect timber development in the future. Past studies have indicated that mechanical terracing increases soil water storage by reducing runoff and retaining more snow, making water more available deeper within the soil profile (Querejeta et al. 2001). In addition, terracing has been shown to increase water infiltration and rooting depth in the subsoil (Ternan et al. 1996, Martínez-Casasnovas et al. 2010). Combining these mechanisms, Gallart et al. (1994) showed that terracing may facilitate soil moisture recharge based on modeled natural hillslope conditions before and after terracing using the TOPMODEL. Given that soil conditions are nearly the same for terrace and nonterrace locations in the BNF, it stands to reason that seasonal snowpack is retained longer on terrace bench locations, reducing runoff and, consequently, increasing infiltration by reducing overland flow. We speculate that this occurred for the terraces on the BNF and resulted in greater dbh growth but was not directly assessed in this study. In addition, encouraging the terrace riser location to establish ground cover helped reduce the amount of erosional soil losses thereby giving the terraces more stability on the mountain slope (Li and Lindstrom 2001). Conclusions and Management Implications Standard and terracing site preparation techniques on the BNF did not express detrimental impacts on timber productivity after 45 years on sites in our study area. Our results show that bulk density among the harvested and terraced plots was not different from unharvested stands illustrating that these soil types have recovered to predisturbance levels. Changes in soil bulk density also did not appear to have had a significant impact on ponderosa pine seedling establishment and growth. Soil disturbance on both mechanically prepared treatments was still detectable but appeared to be recovering from the previous disturbances. However, there was reduced understory vegetation cover, species diversity, and reduced forest floor biomass and OM content on the mechanically treated locations. Terrace construction created more horizontal surface area along the contour of the hill slope, benefiting young ponderosa pine seedlings by offering a larger soil volume to establish rooting systems on otherwise marginal lands. Most of the mechanical site preparation areas had low nutrients or low OM soils. Site preparation on these particular soil types may have enhanced these marginal conditions and given plantation trees a competitive advantage. In addition, the high rock content of these soils may have buffered the soil fine-fraction from compaction and were less impacted than finer textured soils. Our findings provide baseline data on soil recovery after extraordinary mechanical site preparation treatments on the BNF. This study can also provide land managers with data for developing best management practices (BMP) for proposed timber harvesting or other land management goals that may require the use of machines on forested land. " We would like to thank Joanne Tirocke and the other employees at the Rocky Mountain Research Station in Moscow, Idaho for all their help and support with lab analysis, statistical consultation, and use of facilities. We are grateful to the Theresa Jain and two anonymous reviewers for helpful comments on the manuscript. This publication was made possible in part by the NSF Idaho EPSCoR Program and by the National Science Foundation under award number EPS-0,814,387. The authors are also grateful to the USDA Forest Service Northern Region for their financial and logistical support of this project. " This article uses metric units; the applicable conversion factors are: centimeters (cm): 1 cm = 0.39 in.; meters (m): 1 m = 3.3 ft; millimeters (mm): 1 mm = 0.039 in.; kilograms (kg); 1 kg = 2.2 lb; megagrams (Mg): 1 Mg = 2,204.6 lb; grams (g) 1 g = 0.035 oz; hectares (ha): 1 ha = 2.47 ac. Literature Cited Ares A. Terry T.A. Miller R.E. Anderson H.W. Flaming B.L. 2005 . Ground-based forest harvesting effects on soil physical properties and Douglas-fir growth . Soil Sci. Soc. Am. J. 69 ( 6 ): 1822 – 1832 . Google Scholar Crossref Search ADS WorldCat Clayton J.L. Kellogg G. Forrester N. 1987 . Soil disturbance-tree growth relations in central Idaho clearcuts . USDA For. Serv., Res. Note INT-372 , Ogden, UT . Doty R.D. 1970 . Influence of contour trenching on snow accumulation . J. Soil Water Conserv. 25 ( 3 ): 102 – 104 . OpenURL Placeholder Text WorldCat Doty R.D. 1972 . Soil water distribution on a contour trenched area . USDA For. Serv., Res. Note RN-INT-163 , Ogden, UT . Gallart F. Llorens P. Latron J. 1994 . Studying the role of old agricultural terraces on runoff generation in a small Mediterranean mountainous basin . J. Hydrol. 159 ( 1–4 ): 291 – 303 . Google Scholar Crossref Search ADS WorldCat Gee G.W. Bauder J.W. 1986 . Particle-size analysis . P. 383 – 411 in Methods of soil analysis, part 1. Physical and mineralogical methods. Am. Soc. Agron., Soil Sci. Soc. Am. , Madison, WI . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Green P. Joy J. Sirucek D. Hann W. Zack A. Naumann B. 1992 . Old-growth forest types of the northern region (errata corrected 2/05) . USDA For. Serv. , Northern Region, Missoula, MT . Homann P.S. Sollins P. Chappell H.N. Stangenberger A.G. 1995 . Soil organic carbon in a mountainous, forested region: Relation to site characteristics . Soil Sci. Soc. Am. J. 59 ( 5 ): 1468 – 1475 . Google Scholar Crossref Search ADS WorldCat Jurgensen M.F. Harvey A.E. Graham R.T. Page-Dumroese D.S. Tonn J.R. Larsen M. J. Jain T.B. 1997 . Impacts of timber harvesting on soil organic matter, nitrogen, productivity, and health of inland northwest forests . For. Sci. 43 : 234 – 251 . OpenURL Placeholder Text WorldCat Karam A. 1993 . Chemical properties of organic soils . P. 459 – 471 in Soil sampling methods and analysis , Carter M.R. (ed.). Canadian Society of Soil Science , Boca Raton, FL . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Kulmatiski A. Vogt D.J. Siccama T.G. Beard K.H. 2003 . Detecting nutrient pool changes in rocky forest soils . Soil Sci. Soc. Am. J. 67 ( 4 ): 1282 – 1286 . Google Scholar Crossref Search ADS WorldCat Li Y. Lindstrom M.J. 2001 . Evaluating soil quality–soil redistribution relationship on terraces and steep hillslope . Soil Sci. Soc. Am. J. 65 ( 5 ): 1500 – 1508 . Google Scholar Crossref Search ADS WorldCat Martínez-Casasnovas J.A. Ramos M.C. Espinal.-Utgés S. 2010 . Hillslope terracing effects on the spatial variability of plant development as assessed by NDVI in vineyards of the Priorat Region (NE Spain) . Environ. Monit. Assess. 163 ( 1–4 ): 379 – 396 . Google Scholar Crossref Search ADS PubMed WorldCat McNabb D.H. Baker-Katz K. Tesch S.D. 1993 . Machine site preparation improves seedling performance on a high-elevation site in south-west Oregon . West. J. Appl. For. 8 : 95 – 98 . OpenURL Placeholder Text WorldCat Mulvaney R.L. 1996 . Nitrogen-inorganic forms . P. 1123 – 1184 in Methods of soil analysis, part 3. Chemical methods , Sparks D.L. (ed.). Soil Science Society of America , Madison, WI . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Nelson D.W. Sommers L.E. 1996 . Total carbon, organic carbon, and organic matter . P. 961 – 1069 in Methods of soil analysis, part 3. Chemical methods , Sparks D.L. (ed.). Soil Science Society of America , Madison, WI . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Newton M. Starker B. Blanchard G. 1974 . Terrace planting in the Oregon Coast Range. Forest Research Lab., School of Forestry, Oregon State University, Res. Pap. 20 . 7 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Nie M. 2008 . The Bitterroot controversy . In Forests and forestry in the Americas: An encyclopedia , Cubbage F.W. (ed.). Routledge , New York . Available online at http://encyclopediaofforestry.org ; last accessed July 10, 2013. NOAA . 2011 . National Oceanic and Atmospheric Administration. National Climatic Data Center. Climate data online . Available online at http://www.ncdc.noaa.gov ; last accessed Mar. 19, 2013. Page-Dumroese D.S. Brown R.E. Jurgensen M.F. Mroz G.D. 1999 . Comparison of methods for determining bulk densities of rocky forest soils . Soil Sci. Soc. Am. J. 63 ( 2 ): 379 – 383 . Google Scholar Crossref Search ADS WorldCat Page-Dumroese D.S. Abbott A.M. Rice T.M. 2009 . Forest soil disturbance monitoring protocol . USDA For. Serv., Gen. Tech. Rep. WO-GTR-82bII , Washington, DC . 64 p. Page-Dumroese D.S. Jurgensen M. Terry T. 2010 . Maintaining soil productivity during forest or biomass-to-energy thinning harvests in the western United States . West. J. Appl. For. 25 ( 1 ): 5 – 11 . OpenURL Placeholder Text WorldCat Page-Dumroese D.S. Abbott A.M. Curran M.P. Jurgensen M.F. 2012 . Validating visual disturbance types and classes used for forest soil monitoring protocols . USDA For. Serv., Gen. Tech. Rep. RMRS-GTR-267 , Ogden, UT . 17 p. Querejeta J.I. Roldan A. Albaladejo J. Castillo V. 2000 . Soil physical properties and moisture content affected by site preparation in the afforestation of a semiarid rangeland . Soil Sci. Am. J. 64 : 2087 – 2096 . Google Scholar Crossref Search ADS WorldCat Querejeta J.I. Roldán A. Albaladejo J. Castillo V. 2001 . Soil water availability improved by site preparation in a Pinus halepensis afforestation under semiarid climate . For. Ecol. Manage. 149 ( 1–3 ): 115 – 128 . Google Scholar Crossref Search ADS WorldCat Ross D.W. Scott W. Heninger R.L. Walstad J.D. 1986 . Effects of site preparation on ponderosa pine (Pinus ponderosa), associated vegetation, and soil properties in south central Oregon . Can. J. For. Res. 16 ( 3 ): 612 – 618 . Google Scholar Crossref Search ADS WorldCat Soil Survey Staff . 2012a . Official soil series descriptions. USDA-NRSC . Available online at http://soils.usda.gov/technical/classification/osd/index.html ; last accessed Mar. 19, 2013. Soil Survey Staff . 2012b . Web soil survey. USDA–NRSC . Available online at http://websoilsurvey.nrcs.usda.gov/ ; last accessed Mar. 19, 2013. Ter-Mikaelian M.T. Korzukhin M.D. 1997 . Biomass equations for sixty-five North American tree species . For. Ecol. Manage. 97 ( 1 ): 1 – 24 . Google Scholar Crossref Search ADS WorldCat Ternan J.L. Williams A.G. Elmes A. Fitzjohn C. 1996 . The effectiveness of bench-terracing and afforestation for erosion control on raña sediments in central Spain . Land Degrad. Develop. 7 ( 4 ): 337 – 351 . Google Scholar Crossref Search ADS WorldCat Thomas G.W. 1982 . Exchangeable cations . P. 159 – 164 in Methods of soil analysis, part 2. Chemical and micro-biological properties , Page A.L. (ed.). Agon. Mongr. 9, Am. Soc. Agron , Madison, WI . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC USDA Forest Service . 1970 . Management practices on the Bitterroot National Forest: A task force appraisal, May 1969–April 1970 , Missoula, MT . Watson A.K. Renny A.J. 1974 . The biology of Canadian weeds 6. Centaurea diffusa and C. maculosa . Can. J. Plant Sci. 54 : 687 – 701 . Google Scholar Crossref Search ADS WorldCat Worf W.A. 1970 . Management practices on the Bitterroot National Forest: A task force analysis, May 1969–April 1970 . USDA For. Serv . OpenURL Placeholder Text WorldCat Zlatnik E.J. De.Luca T.H. Milner K.S. Potts D.F. 1999 . Site productivity and soil conditions on terraced ponderosa pine sites in western Montana . West. J. Appl. For. 14 ( 1 ): 35 – 40 . OpenURL Placeholder Text WorldCat Copyright © 2013 by the Society of American Foresters.

Cook, Stephen, P.;Sloniker, Benjamin, D.;Rust, Marc, L.

doi: 10.5849/wjaf.13-002pmid: N/A

One management technique that has been receiving attention for use against various forest insects are applications of bole-injected systemic insecticides. We evaluated the use of two such insecticides (emamectin benzoate applied at a rate of 0.07 ml AI/cm dbh and imidacloprid applied at a rate of 0.1 ml active ingredient (AI)/cm dbh) applied at two different times of the year (fall versus spring) for management of Douglas-fir tussock moth, Orgyia pseudotsugata (McDunnough) (Lepidoptera: Lymantriidae) and the fir coneworm, Dyorictria abietivorella (Grote) (Lepidoptera: Pyralidae), insects that cause damage to Douglas-fir foliage and cones, respectively. Spring and fall bole injection of emamectin benzoate significantly reduced survival of Douglas-fir tussock moth caterpillars fed foliage from treated trees and reduced fir coneworm infestations on treated trees compared with untreated controls. Both fall and spring injection periods were equally effective. Douglas-fir tussock moth caterpillar survival was also significantly lowered when they fed on foliage from trees that had received spring injections of imidacloprid. At the dose tested, the fall treatment of imidacloprid did not reduce Douglas-fir tussock moth survival and neither fall or spring treatments significantly reduced infestation levels of fir coneworm. The results suggest that bole-injection treatments of emamectin benzoate can be effective in minimizing insect damage when applied the year prior to foliar feeding or harvest of the cone crop. Imidacloprid appears to be effective only for foliage protection and only when applied during the spring of the year that protection is needed. Orgyia pseudotsugata, Dioryctria abietivorella, Pseudotsuga menziesii, emamectin benzoate, imidacloprid Two common insect pests of Douglas-fir, Pseudotsuga menziesii (Mirb.) Franco, in the Pacific Northwest are the Douglas-fir tussock moth, Orgyia pseudotsugata (McDunnough) (Lepidoptera: Lymantriidae) and the fir coneworm, Dyorictria abietivorella (Grote) (Lepidoptera: Pyralidae). Douglas-fir tussock moth is a univoltine defoliator that primarily attacks true firs as well as Douglas-fir (Furniss and Carolin 1977). The moth occurs from British Columbia southward to Arizona and New Mexico and from the Pacific coast eastward to Montana, Wyoming, and Colorado (Wickman et al. 1981). Although outbreaks develop quickly and subside after a year or two, defoliation by the caterpillars can result in tree mortality, topkill of defoliated trees, or reduced tree vigor that can render them more susceptible to other mortality agents. There is a complex of natural enemies that use Douglas-fir tussock moth, including a naturally occurring virus (Vezina and Peterman 1985), along with several predators (Mason et al. 1983) and parasitoids (Torgersen 1977, Cook et al. 2005). These natural enemies typically bring population outbreaks under control but other management techniques such as the use of aerially applied insecticides (i.e., McGrath 2001, Cook 2003) are used. Fir coneworm is a common cone pest in seed orchards producing Douglas-fir, pine, and other conifer seeds and infestations can result in high levels of damage, including up to almost complete destruction of the cone crop during some years (Hedlin et al. 1981). Fir coneworm can also cause damage by mining buds, shoots, and trunks of trees (Furniss and Carolin 1977). Conifer seed orchards are frequently managed to minimize damage from fir coneworm due to the potentially high levels of seed loss caused by feeding damage. The management techniques used for fir coneworm typically include the application of insecticides using ground-based hydraulic sprayers (DeBarr et al. 1982, Haverty et al. 1986, Rappaport et al. 1994). Bole-injected systemic insecticides have been tested against Dioryctria coneworms in two pine systems, slash pine (Pinus elliottii Engelmann) in the southeastern United States (Merkel 1970) and ponderosa pine (P. ponderosa (Laws)) in Idaho (Cook et al. 2013). Bole-injected systemic insecticides have been demonstrated to be useful for management of a variety of conifer-infesting insects. While not all systemic insecticide treatments have proven to be effective against conifer-infesting bark beetles (Fettig et al. 2010), treatments have been effective against several species of bark beetles (Coleoptera: Curculionidae) (DeGomez et al. 2006, Grosman and Upton 2006, Grosman et al. 2009, 2010), as well as other conifer pests such as hemlock woolly adelgid, Adelges tsugae (Hemiptera: Adelgidae) attacking eastern hemlock (i.e., Cowles et al. 2006), ponderosa pine cone beetle, Conophthorus ponderosae Hopkins (Coleoptera: Curculionidae) (Cook et al. 2013) and Dioryctria coneworms in pine (Merkel 1970, Cook et al. 2013). When a compound such as a systemic insecticide is injected into the bole of a tree, it requires time to be transported to various tree tissues. Insects that attack and cause damage early in the growing season or to developing tissues such as current year's foliage may require fall treatments to ensure that the insecticide has translocated to feeding sites prior to feeding initiation in the spring. Egg hatch of Douglas-fir tussock moth occurs at approximately the same time as bud break. Caterpillars initially feed on new foliage and switch to older foliage as they mature (Wickman et al. 1981). Fir coneworm may have a partial second generation during a year with adults emerging as early as May or June (Hedlin et al.1981). In a prior experiment, we examined the effectiveness of two bole-injected systemic insecticides, emamectin benzoate and imidacloprid, on two common pests in ponderosa pine seed orchards (Cook et al. 2013). Both fall and spring treatments of emamectin benzoate decreased damage caused by the ponderosa pine cone beetle. There was no difference in the effectiveness of fall versus spring imidacloprid treatments on coneworms but only the spring treatment resulted in reduced damage compared with the untreated controls. The objective of the current study was to compare the efficacy of spring versus fall treatments of the bole-injected, systemic insecticides emamectin benzoate and imidacloprid at reducing damage caused by Douglas-fir tussock moth and fir coneworm. Methods Field Sites and Experimental Treatments Fieldwork was conducted at the Idaho Department of Lands, Paradise Valley Douglas-fir seed orchard located approximately 6.5 km southwest of Bonners Ferry, ID (latitude = 48.6,43,227; longitude = −116.3,46,356). The seed orchard was established from planted seedlings in 1985. Potential study trees were selected based on bole diameter at 1.4 m (dbh). Only tree diameter was used for comparing tree size because trees within the orchard are periodically topped to control height and all trees had similar heights. Trees within the orchard are approximately 8 m apart. All of the study trees were injected with gibberellins (GA4 + 7) (ProCone, Valent BioSciences Corporation) during late June 2011 to stimulate cone production for 2012. Trees were randomly assigned to one of five treatments: controls that received no injection, fall injections of either imidacloprid or emamectin benzoate, and spring injections of either imidacloprid or emamectin benzoate. Each treatment was replicated 10 times. Trees did not differ in size among the treatment groups (F = 1.47; df = 4, 45; [P > F] = 0.2275). The mean (± standard error of the mean [SEM]) dbh for the trees in each treatment group were: nontreated controls = 30.5 ± 1.3 cm; imidacloprid, fall treatment = 30.2 ± 1.5 cm; imidacloprid, spring = 31.5 ± 1.0 cm; emamectin benzoate, fall = 32.4 ± 1.1 cm; and emamectin benzoate, spring = 33.7 ± 1.0 cm. Treatments were applied to the appropriate trees during either a fall, 2011 treatment period (November 2) or spring, 2012 treatment period (May 9). The imidacloprid formulation used for the test was IMA-jet (5.0% AI solution) and the emamectin benzoate formulation used was TREE-age (4.0% AI solution). The compounds were applied using a Quick-jet microinjection system (ARBORjet, Inc., Woburn, MA). Imidicloprid was applied at a dose of 0.1 ml active ingredient (AI) per cm dbh and emamectin benzoate was applied at a dose of 0.07 ml AI per cm dbh. Treatments were injected into the trees after drilling holes at a height of approximately 50 cm aboveground level and placing Arborplugs (Arborjet, Inc.) into the hole opening. Holes were drilled at approximately 45° angles, had a diameter of approximately 0.9 cm, and were approximately 6 cm deep. Depending on the diameter of the individual trees, there were from four to seven evenly spaced holes per stem. Foliage from all of the trees was used for the evaluation of Douglas-fir tussock moth caterpillar survival. However, a minimum of 20 cones per tree was set as a threshold for evaluating damage by fir coneworm to limit potential bias caused by low-producing trees. While the threshold was predetermined to be 20 cones per tree, the trees eliminated from the study had 10 or fewer cones visible at the time data were collected. Because of this threshold, not all trees were included in the estimates of fir coneworm infestations (n = 6 control trees, 4 fall-injected emamectin benzoate trees, 5 spring-injected emamectin benzoate trees, 8 fall-injected imidacloprid trees, and 2 spring-injected imidacloprid trees). Douglas-Fir Tussock Moth Survival Douglas-fir tussock moth egg masses were collected from an infestation near Plummer, ID in October 2011. Egg masses were surface sterilized to eliminate contamination with the nuclear polyhedrosis virus that is frequently associated with Douglas-fir tussock moth infestations (Vezina and Peterman 1985). Egg masses were maintained at 0° C until May 2012. Newly hatched caterpillars were placed on synthetic diet (BioServ) on May 28, 2012. Caterpillars were placed 20 per cup in 500 ml cups, approximately 25% filled with diet and maintained at 22° C in a 12 hour light: 12 hour dark regimen. Second instar caterpillars were placed on foliage from treated and untreated trees on June 13 (1 l glass containers with mesh tops) and maintained at 22° C. Only foliage from the current year was used for the study. Foliage was collected by clipping branches from the midcanopy of trees on June 12. Foliage was kept hydrated by placing individual cut branches in sealed glass vials throughout the feeding period. Five branch tips with foliage were placed into the glass containers with the caterpillars. There was not enough feeding to require the addition of more foliage. Caterpillar survival was measured on June 20 and 27. Surviving caterpillars were transferred to a synthetic diet on June 27, one caterpillar per 30 ml cup, approximately 25% filled with diet and maintained at 22° C. The transfer of caterpillars to synthetic diet allowed us to determine if additional mortality would occur even if caterpillars switched from a treated to an untreated tree. Caterpillar survival was recorded on July 4 and 11. Data were tested for homogeneity of variance and nontransformed data were compared using analysis of variance (ANOVA) tests, ANOVA (Analytical Software 2009). Comparisons were conducted on the percentage survival of caterpillars among treatments after feeding for 5 weeks on foliage from treated and untreated trees (June 27) and 2 weeks following their transfer to synthetic diet (July 11). Fir Coneworm Damage Damage to cones was evaluated on Aug. 2, 2012. The date was selected because early August coincides with the approximate timing of when a harvest would occur. Fir coneworm damage includes the deposition of frass and webbing on the external surface of the cones and is conspicuous (Hedlin et al. 1981). Further, there had not been any rain events to disturb the frass on the cone surface for two weeks prior to the measurements. Fir coneworm damage was assessed by two observers. Observers stood on opposite aspects of a tree and, using binoculars, each evaluated up to 50 cones per tree, starting at the top of the crown and working down. Multiple angles were used to see as much of the cone surface as possible. Cones were counted as infested by fir coneworm based on the presence of frass on the external surface of the cones (Hedlin et al. 1981). Using this technique, any bias would be toward recording a cone as healthy versus infested because frass could have been present on the bole-facing aspect of a cone and not observable. However, observations were made from multiple points, which decreased the possibility of this bias occurring. Data were tested for homogeneity of variance and nontransformed data were compared using ANOVA tests to determine differences among treatments in the overall percentage of healthy cones per tree. Results Douglas-Fir Tussock Moth Douglas-fir tussock moth caterpillars that were initially reared on foliage from trees treated with bole-injected systemic insecticides had lower survival rates over the 4-week feeding period compared with caterpillars reared on foliage from nontreated trees (Figure 1). There were significantly fewer caterpillars that survived on foliage from trees treated with emamectin benzoate (both fall and spring treatments) or trees treated during the spring with imidacloprid at both 2 weeks (F = 3.90; df = 4, 45; [P > F] ≤ 0.0083) and 4 weeks (F = 4.36; df = 4, 45; [P > F] ≤ 0.0046) (Table 1). However, caterpillar survival on untreated foliage was not significantly different from fall imidacloprid treated foliage over the 4-week period. Figure 1. Open in new tabDownload slide Mean percentage survival of Douglas-fir tussock moth caterpillars for four weeks after being transferred from synthetic diet and fed for 2 weeks on foliage from Douglas-fir trees that was removed on June 12, 2012 and then feeding on synthetic diet for an additional 2 weeks. Trees had received bole-injection treatments of imidacloprid (IM) or emamectin benzoate (EB) applied on Nov. 2, 2011 (fall) or on May 9, 2012 (spring). Figure 1. Open in new tabDownload slide Mean percentage survival of Douglas-fir tussock moth caterpillars for four weeks after being transferred from synthetic diet and fed for 2 weeks on foliage from Douglas-fir trees that was removed on June 12, 2012 and then feeding on synthetic diet for an additional 2 weeks. Trees had received bole-injection treatments of imidacloprid (IM) or emamectin benzoate (EB) applied on Nov. 2, 2011 (fall) or on May 9, 2012 (spring). Table 1. Mean percentage survival (± SEM) of Douglas-fir tussock moth caterpillars at 2 and 4 weeks following placement on treated foliage. Caterpillars fed for 2 weeks on foliage removed from Douglas-fir trees on June 12, 2012 and then fed on synthetic diet for 2 weeks. Trees received bole-injection treatments of emamectin benzoate (EB) or imidacloprid (IM) applied on Oct. 17, 2011 (fall) or on May 9, 2012 (spring). a Based upon analysis of variance tests, within a column, means followed by the same letter are not significantly different (α = 0.05). Open in new tab Table 1. Mean percentage survival (± SEM) of Douglas-fir tussock moth caterpillars at 2 and 4 weeks following placement on treated foliage. Caterpillars fed for 2 weeks on foliage removed from Douglas-fir trees on June 12, 2012 and then fed on synthetic diet for 2 weeks. Trees received bole-injection treatments of emamectin benzoate (EB) or imidacloprid (IM) applied on Oct. 17, 2011 (fall) or on May 9, 2012 (spring). a Based upon analysis of variance tests, within a column, means followed by the same letter are not significantly different (α = 0.05). Open in new tab Fir Coneworm Significantly fewer cones (F = 6.30; df = 4, 20; [P > F] ≤ 0.0019) were infested with fir coneworm on the Douglas-firs that received either a spring or fall treatment with emamectin benzoate compared with untreated controls (Table 2). Although significantly different from the spring emamectin benzoate treatment, cone damage in trees that were treated with imidacloprid injected during the spring were not significantly different from the damage on trees that received the fall emamectin benzoate treatment. The cone damage on trees treated during the fall or spring with imidacloprid did not differ significantly from the cone damage on nontreated control trees. Table 2. Mean percentage of cones (± SEM) not infested with fir coneworm at the end of the growing season (Aug. 2, 2012) from trees that had received bole-injection treatments of imidacloprid (IM) or emamectin benzoate (EB) applied on Oct. 17, 2011 (fall) or on May 9, 2012 (spring). a Based upon analysis of variance tests, within a column, means followed by the same letter are not significantly different (α = 0.05). Open in new tab Table 2. Mean percentage of cones (± SEM) not infested with fir coneworm at the end of the growing season (Aug. 2, 2012) from trees that had received bole-injection treatments of imidacloprid (IM) or emamectin benzoate (EB) applied on Oct. 17, 2011 (fall) or on May 9, 2012 (spring). a Based upon analysis of variance tests, within a column, means followed by the same letter are not significantly different (α = 0.05). Open in new tab Discussion At the dose used, bole injections of emamectin benzoate significantly reduced survival of Douglas-fir tussock moth caterpillars after 2 weeks of feeding/exposure. Caterpillars continued to die (an additional 3 and 13% in the fall and spring treatments, respectively) during the 2-week period following removal from the treated foliage. Caterpillars fed on foliage from control trees had a 78% survival rate compared to a 35% survival of caterpillars fed on foliage from trees that received either a fall or spring treatment of emamectin benzoate. In a prior study (Cook et al. 2013), emamectin benzoate was effective against both the ponderosa pine cone beetle and coneworms in ponderosa pine. These results suggest protection of trees from a variety of insects can be achieved with a fall treatment. Therefore, accessibility problems that may be associated with implementing treatments in the early spring can be avoided. Although there was no statistically significant difference between the fall and spring treatment periods, it did appear to take longer for the mortality to occur for caterpillars feeding on foliage from the spring- versus fall-treated trees. Trees that were treated during the spring injection period may not have had time to completely move the material up the bole and into the foliage. If the emamectin benzoate was in lower concentrations in the foliage it may have taken longer to reach a dose that would result in caterpillar death. However, the caterpillars were moved after 2 weeks to a nontreated synthetic diet. Therefore, they would have had to acquire a lethal dose during the 2 weeks that they were feeding on foliage from the treated trees. While the fall injection of imidacloprid did not decrease caterpillar survival, there was a significant reduction in survival of the Douglas-fir tussock moth caterpillars that fed on foliage from trees that had received a spring injection of imidacloprid. Similar to the results with emamectin benzoate, mortality of the caterpillars took longer if they were fed foliage from spring-treated trees. Fir coneworm damage was lowest in the trees that had received either a fall or spring bole-injection treatment of emamectin benzoate. Both fall and spring treatments with emamectin benzoate also reduced cone damage in ponderosa pine (Cook et al. 2013) and spring treatments provided multiple year protection of loblolly pine cones from coneworm damage (Grosman et al. 2002). This study and others indicate bole injection treatments of emamectin benzoate applied in the fall limit cone damage the year following treatment. During the current study, trees that received a spring treatment with imidacloprid had the highest level of damage from fir coneworm. Further, the cone damage that occurred in trees that had received either the fall or spring treatments of imidacloprid was not significantly different from the damage on untreated trees. One problem with our data is that sample sizes for this portion of the study were limited due to low cone numbers on some trees. Previous studies indicate spring treatments of imidacloprid reduced coneworm damage in ponderosa pine (Cook et al. 2013) and in loblolly pine (Grosman et al. 2002). The current study and others suggest that bole-injection treatments of imidacloprid may be effective at reducing coneworm damage if it is applied during the spring after dose and timing questions are addressed. " We thank the Idaho Department of Lands for allowing us to conduct the work at the Paradise Valley Douglas-fir seed orchard. We also thank Lindsay Menard and William Sweeney who helped with data collection. The work was funded in part by grants from the National Science Foundation-Center for Advanced Forestry Systems, the USDA Forest Service, Pesticide Impact Assessment Program, and the Inland Empire Tree Improvement Cooperative. This article reports the results of research only. Mention of a proprietary product does not constitute an endorsement or recommendation for its use. " This article uses metric units; the applicable conversion factors are: centimeters (cm): 1 cm = 0.39 in.; meters (m): 1 m = 3.3 ft; kilometers (km): 1 km = 0.6 mi. Literature Cited Analytical Software . 2009 . Statistix 8 user's manual. Tallahassee, FL . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Cook S.P. 2003 . Laboratory and field evaluation of tebufenozide, diflubenzuron, and Bacillus thurengiensis var. kurstaki for suppression of Douglas-fir tussock moth (Orgyia pseudotsugata (McDunnough)) in Idaho: A case study . J. Econ. Entomol. 96 : 396 – 400 . Google Scholar Crossref Search ADS PubMed WorldCat Cook S.P. Sloniker B.D. Rust M.L. 2013 . Using systemically-applied insecticides for management of ponderosa pine cone beetle in pine seed orchards . West. J. Appl. For. 28 : 66 – 70 . Google Scholar Crossref Search ADS WorldCat Cook S. Wenz J. Ragenovich I. Reardon R. Randall C. 2005 . Impact of mating disruption pheromone treatments to control Douglas-fir tussock moth, on egg parasitoids . Pan-Pacific Entomol. 81 : 41 – 46 . OpenURL Placeholder Text WorldCat Cowles R.S. Montgomery M.E. Cheah C.A.S.-J. 2006 . Activity and residues of imidacloprid applied to soil and tree trunks to control hemlock woolly adelgid (Hemiptera: Adelgidae) in forests . J. Econ. Entomol. 99 : 1258 – 1267 . Google Scholar Crossref Search ADS PubMed WorldCat DeBarr G.L. Barber L.R. Maxwell A.H. 1982 . Use of carbofuran for control of eastern white pine cone and seed insects . For. Ecol. Manage. 4 : 1 – 18 . Google Scholar Crossref Search ADS WorldCat DeGomez T.E. Hayes C.J. Anhold J.A. McMillin J.D. Clancy K.M. Bosu P.P. 2006 . Evaluation of insecticides for protecting southwestern ponderosa pine from attack by engraver beetles (Coleoptera: Curculionidae: Scolytinae) . J. Econ. Entomol. 99 : 393 – 400 . Google Scholar Crossref Search ADS PubMed WorldCat Fettig C.J. Munson A.S. Jorgensen C.L. Grosman D.M. 2010 . Efficacy of fipronil for protecting individual pines from mortality attributed to attack by western pine beetle and mountain pine beetle (Coleoptera: Curculionidae, Scolytinae) . J. Entomol. Sci. 45 : 296 – 301 . Google Scholar Crossref Search ADS WorldCat Furniss R.L. Carolin V.M. 1977 . Western forest insects. USDA For. Serv., Misc. Publ. No. 1339 . 654 p. Grosman D.M. Clarke S.R. Upton W.W. 2009 . Efficacy of two systemic insecticides injected into loblolly pine for protection against southern pine bark beetles (Coleoptera: Scolytidae) . J. Econ. Entomol. 102 : 1062 – 1069 . Google Scholar Crossref Search ADS PubMed WorldCat Grosman D.M. Fettig C.J. Jorgensen C.L. Munson A.S. 2010 . Effectiveness of two systemic insecticides for protecting western conifers from mortality due to bark beetle attack . West. J. Appl. For. 25 : 181 – 185 . OpenURL Placeholder Text WorldCat Grosman D.M. Upton W.W. 2006 . Efficacy of systemic insecticides for protection of loblolly pine against southern pine engraver beetles (Coleoptera: Curculionidae: Scolytidae) and wood borers (Coleoptera: Cerambycidae) . J. Econ. Entomol. 99 : 94 – 101 . Google Scholar Crossref Search ADS PubMed WorldCat Grosman D.M. Upton W.W. McCook F.A. Billings R.F. 2002 . Systemic insecticide injections for control of cone and seed insects in loblolly pine seed orchards–2 year results . South. J. Appl. For. 26 : 146 – 152 . OpenURL Placeholder Text WorldCat Haverty M.I. Shea P.J. Stipe L.E. 1986 . Single and multiple applications of fenvalerate to protect cones from Dioryctria abietivorella (Lepidoptera: Pyralidae) . J. Econ. Entomol. 79 : 158 – 161 . Google Scholar Crossref Search ADS WorldCat Hedlin A.F. Yates H.O. III Cibrian-Tovar D. Ebel B.H. Koerber T.W. Merkel E.P. 1981 . Cone and seed insects of North American conifers. Can. For. Serv., USDA For. Serv., Secretaria de Agricultura y Recursos hidraulicos , Mexico . 122 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mason R.R. Torgersen T.R. Wickman B.E. Paul H.G. 1983 . Natural regulation of a Douglas-fir tussock moth (Lepidoptera: Lymantriidae) population in the Sierra Nevada . Environ. Entomol. 12 : 587 – 594 . Google Scholar Crossref Search ADS WorldCat McGrath D. 2001 . 2001 Pacific Northwest insect management handbook. Oregon State University , Corvallis, OR . 660 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Merkel E.P. 1970 . Trunk-implanted systemic insecticides for slash pine cone insect control . Florida Entomol. 53 : 143 – 146 . Google Scholar Crossref Search ADS WorldCat Rappaport N.G. Haverty M.J. Shea P.J. Sandquist R.E. 1994 . Efficacy of esfenvalerate for control of insects harmful to seed production in disease-resistant western white pines . Can. Entomol. 126 : 1 – 5 . Google Scholar Crossref Search ADS WorldCat Torgersen T.R. 1977 . Identification of the parasites of the Douglas-fir tussock moth, based on adults, cocoons, and puparia. USDA For. Serv., Res. Pap. PNW-215 . 28 p. OpenURL Placeholder Text WorldCat Vezina A. Peterman R.M. 1985 . Tests of the role of a nuclear polyhedrosis virus in the population dynamics of its host, Douglas-fir tussock moth, Orgyia pseudotsugata (Lepidoptera: Lymantriidae) . Oecologia 67 : 260 – 266 . Google Scholar Crossref Search ADS PubMed WorldCat Wickman B.E. Mason R.R. Trostle G.C. 1981 . Douglas-fir tussock moth. USDA For. Serv., Forest Insect and Disease Leaflet 86 . 10 p. OpenURL Placeholder Text WorldCat Copyright © 2013 by the Society of American Foresters.

Fidgen, Jeffrey, G.;Kittelson, Neal, T.;Eckberg,, Tom;Doccola,, Joe;Randall,, Carol

doi: 10.5849/wjaf.12-036pmid: N/A

Stem injections of insecticides are generally regarded as a safer and a more environmentally friendly option (compared to foliar sprays) for protecting high-value trees against insects in sensitive areas or near homes. We carried out a three year study to determine the efficacy of trunk injections of emamectin benzoate for protection of foliage of three common host species attacked by the western spruce budworm, Choristoneura occidentalis Freeman (Lepidoptera: Tortricidae). Defoliation by C. occidentalis was significantly reduced by nearly half over a 3-year period on treated compared to control grand fir, Douglas-fir, and alpine fir. Reduction of defoliation was comparable to standards used for aerial spraying (e.g., 50%) of conifer feeding budworms in Canada. western spruce budworm, stem injection, TREE-äge, polyphagous, budworm The western spruce budworm, Choristoneura occidentalis Freeman, is a native, deleterious pest of the forests of the Pacific Northwest and the intermountain and northern regions of the United States. In 2009, approximately 1.2 million ha of the 1.7 million ha of defoliation caused by C. occidentalis reported among these regions occurred in the northern region, specifically the Idaho panhandle and adjacent areas in western Montana.1 This problem is exacerbated by the lack of suppression of high intensity C. occidentalis populations on national forest lands. The western spruce budworm feeds on spruce (Picea), true firs (Abies), and Douglas-fir, Pseudotsuga menziesii (Mirb.) and occasionally on western hemlock, Tsuga heterophylla (Raf.) Sarg., and western larch, Larix occidentalis Nutt. (Furniss and Carolin 1977). In the spring, the larvae of the budworm feed preferentially on the buds but will feed on male and female cones when available as well as new needles. Heavy feeding by C. occidentalis can significantly reduce photosynthetic capacity and growth (Kramer and Kozlowski 1979, Crookston 1985). Prolonged defoliation by C. occidentalis may cause top kill of trees and sometimes tree death (Crookston 1985, Mason et al. 1997). Heavily defoliated trees may be at greater risk of attack by bark beetles like the Douglas-fir beetle, Dendroctonus pseudotsugae Hopkins, or fir engraver, Scolytus ventralis (LeConte) (Wright and Berryman 1978). As the wildland-urban interface continues to expand in the United States, new homes will be built near hosts susceptible to C. occidentalis (Kline et al. 2004). As human settlements encroach into wilder lands, the inhabitants may put themselves at risk of insecticide exposure due to the close proximity of their homes to forests that may require aerial treatment. In such situations, aerial spraying of trees may be unpopular and expensive due to proximity of trees to homes. Untreated trees with repeated heavy defoliation could die back and become dangerous to nearby homes and require costly removal to mitigate safety risks. Because of the inherent risks of spray drift from aerial spray applications, alternative forms of insecticide treatment, such as stem injections, may protect high-value trees near homes (Cowles et al. 2006, Fry et al. 2008, Grosman et al. 2009, Mota-Sanchez et al. 2009), providing a less risky option for homeowners wishing to maintain and protect their trees. Emamectin benzoate (Syngenta Crop Protection, Greensboro, NC) (hereafter emamectin) has been demonstrated to control several Lepidopteran pests of agricultural crops and protects foliage from various insect pests when injected into trees (e.g., Argentine et al. 2002, Grosman et al. 2002, Ioriatti et al. 2008, Potter et al. 2005, Grosman et al. 2009, Sial and Brunner 2010). Emamectin is a semisynthetic derivative of abamectin, which is itself derived from avermectin through fermentation of the soil bacterium Streptomyces avermitilis (Poland et al. 2006). Emamectin is very immobile in soil (O'Grodnick et al. 1998a) and will not bioaccumulate or magnify in the food chain (Chukwudebe et al. 1996, Mushtaq et al. 1997) or cause adverse effects to large birds even at high concentrations (e.g., 125 ppm in feed) (O'Grodnick et al. 1998b). Stem injections require less active ingredient per tree compared to soil and foliar spray applications of the same compound and reduce exposure to nontarget organisms at the time of application. A relatively new insecticide formulation of emamectin, TREE-äge, has recently been formulated for use in trees (Arborjet, Inc. 2010) and has been tested on several forest insect pests (e.g., Grosman et al. 2002, 2009, Poland et al. 2006, Ioriatti et al. 2009, Sial and Brunner 2010). This formulation has not been tested for efficacy against C. occidentalis. If emamectin provides suitable protection of foliage of hosts susceptible to C. occidentalis, it could be added to integrated pest management programs for this pest. We evaluated the effect of emamectin on defoliation of three tree species commonly attacked by C. occidentalis over a 3-year period following a single trunk injection. The three species we evaluated were Douglas-fir, grand fir, and alpine fir. For treatment to be considered successful, we required that defoliation of emamectin-treated trees had to be 50% or more lower than defoliation of control trees. This level of reduction is targeted to confirm successful suppression of major defoliating insects such as spruce budworm, C. fumiferana Clemens (SOPFIM 2012). Methods Three stands of trees, all located within the Coeur d'Alene National Forest east of Coeur d'Alene, ID, were selected to test the effect of emamectin on defoliation by C. occidentalis, as measured by defoliation of current year shoots. Injections were carried out at a stand (1) near Leiberg Saddle; a second stand approximately 2 km east of Maggie Airstrip; and a third stand approximately 5 km northwest of Maggie Airstrip (Table 1). To our knowledge, none of these locations had been treated with any insecticides during the last 20 years. Table 1. Location and characteristics of stands as well as species injected at each location during 2006 (stand 1) and 2008 (stands 2 and 3). * Injections were carried out Aug. 28, 2006. † Injections were carried out June 3. Weather conditions at the time of June injection prevented all trees being injected; the remaining trees were injected July 9. ‡ Injections were carried out Aug. 1, 2008. # Df = Douglas-fir; Gf = grand fir; wLa = western larch; Wwp = western white pine; Af = alpine fir. Open in new tab Table 1. Location and characteristics of stands as well as species injected at each location during 2006 (stand 1) and 2008 (stands 2 and 3). * Injections were carried out Aug. 28, 2006. † Injections were carried out June 3. Weather conditions at the time of June injection prevented all trees being injected; the remaining trees were injected July 9. ‡ Injections were carried out Aug. 1, 2008. # Df = Douglas-fir; Gf = grand fir; wLa = western larch; Wwp = western white pine; Af = alpine fir. Open in new tab Tree selection was based on accessibility of lower crown branches for sampling to estimate defoliation by C. occidentalis and on visual confirmation, with binoculars, of at least moderate levels of defoliation (e.g., >25% missing needles on new shoots). We selected a total of 10 grand firs and 40 grand firs and 40 Douglas-firs, and 40 alpine firs in stands 1, 2, and 3, respectively, for the experiments. Specifically, five pairs of grand fir trees (one treatment and one control) were haphazardly selected in stand 1 in 2006 with at least 20 m between pairs. Twenty pairs of grand fir and 20 pairs of Douglas-fir trees (one treatment and one control tree per species) were haphazardly selected in stand 2. At stand 3, we selected 20 pairs of alpine fir. Tree pairs at stands 2 and 3 were separated by at least 10 m in the event of grafted root systems. For each pair, one tree was randomly assigned treatment with emamectin and the other tree left alone as a control. In 2006, in stand 1, trees were injected with a 4% active ingredient (AI) formulation of emamectin at a rate of 4 ml/cm dbh using the Tree I.V. system (Arborjet, Inc., Woburn, MA). In 2008, in stands 2 and 3, we switched to a newer injection system, the QUIK-jet and used the same solution of emamectin described above. Injection procedures were carried out in accordance with the emamectin label. Previous budworm defoliation was very light at stand 3, so we waited until budworm feeding had finished before selecting the most defoliated trees (Table 1). During August 2007 and 2008, approximately 1 and 2 years posttreatment of trees at stand 1, we cut a 45 cm branch tip from each cardinal direction in the lower half of the live crown. We determined the percent of missing needles on the 25 most distal current year shoots of each branch tip, with each shoot assessed in 10% increments of missing foliage (Fettes 1950). To evaluate the effect of emamectin on defoliation by C. occidentalis at 1, 2, and 3 years posttreatment, the same collections of branch tips, as described for stand 1, were made from 2009 to 2011 from the trees in stands 2 and 3. Only the 10 most distal current year shoots were assessed on each of four branch tips. For all data, two-way analyses of variance were used to determine if a significant proportion of the variation in defoliation was attributable to the year × treatment interaction separately for the grand fir at stand 1 and for each tree species injected at stands 2 and 3 in 2008 (Table 1). In addition, if defoliation of treated trees was reduced by half or more compared to control trees, we considered efficacy comparable to aerial application of insecticides for control of C. occidentalis. Data are presented as means ± standard errors and were considered significantly different at the 5% probability level. All analyses were performed in the R statistical package (v. 2.11.1, MASS package) on arcsine-square root transformed data, followed by Tukey's tests used to separate significant means at P < 0.05 or less (R Development Core Team 2010). Results At stand 1, a significant proportion of the variation in defoliation was not explained by the year × treatment interaction (Table 2). However, both year and treatment explained significant proportions of the variation in defoliation by C. occidentalis (Table 2). Defoliation was least on treated trees and defoliation in 2008 was significantly higher than 2007 levels (Figure 1). Defoliation of treated trees was reduced by greater than half that of untreated trees in 2007 and 2008. Table 2. Two-way analysis of variance evaluating the effects of year (posttreatment) and treatment on defoliation of grand fir by C. occidentalis at stand 1. Injections were carried out in 2006 and shoot sampling was done in fall 2007 and 2008. Open in new tab Table 2. Two-way analysis of variance evaluating the effects of year (posttreatment) and treatment on defoliation of grand fir by C. occidentalis at stand 1. Injections were carried out in 2006 and shoot sampling was done in fall 2007 and 2008. Open in new tab Figure 1. Open in new tabDownload slide Influence of treatment on defoliation of grand fir (% ± SE) by C. occidentalis at stand 1 in 2007 and 2008. Figure 1. Open in new tabDownload slide Influence of treatment on defoliation of grand fir (% ± SE) by C. occidentalis at stand 1 in 2007 and 2008. A significant proportion of the variation in defoliation was explained by the year × treatment interaction for grand fir and Douglas-fir at stand 2 and for alpine fir at stand 3 (Tables 3 and 4). Defoliation increased on grand fir and Douglas-fir whereas it decreased on alpine fir over the 3 years of the study (Figure 2). Defoliation on treated grand fir was reduced by more than half that of untreated trees in 2009, 2010, and 2011 (Figure 2). For Douglas-fir, defoliation on treated trees was reduced by more than half that of untreated trees in 2009 but was only 48 and 45% lower than levels on untreated trees in 2010 and 2011 (Figure 2). Defoliation on alpine fir was 54, 43, and 48% lower than defoliation on untreated trees in 2009, 2010, and 2011 (Figure 2). Table 3. Two-way analysis of variance evaluating the effects of year (posttreatment) and treatment on defoliation of grand fir and Douglas-fir by C. occidentalis at stand 2. Injections were carried out in 2008 and shoot sampling was done in fall 2009, 2010, and 2011. Open in new tab Table 3. Two-way analysis of variance evaluating the effects of year (posttreatment) and treatment on defoliation of grand fir and Douglas-fir by C. occidentalis at stand 2. Injections were carried out in 2008 and shoot sampling was done in fall 2009, 2010, and 2011. Open in new tab Table 4. Two-way analysis of variance evaluating the effects of year (posttreatment) and treatment on defoliation of alpine fir by C. occidentalis at stand 3. Injections were carried out in 2008 and shoot sampling was done in fall 2009, 2010, and 2011. Open in new tab Table 4. Two-way analysis of variance evaluating the effects of year (posttreatment) and treatment on defoliation of alpine fir by C. occidentalis at stand 3. Injections were carried out in 2008 and shoot sampling was done in fall 2009, 2010, and 2011. Open in new tab Figure 2. Open in new tabDownload slide Influence of treatment on defoliation (% ± SE) by C. occidentalis on grand fir and Douglas-fir at stand 2 and alpine fir at stand 3 from 2009 to 2011. See text for further details. Figure 2. Open in new tabDownload slide Influence of treatment on defoliation (% ± SE) by C. occidentalis on grand fir and Douglas-fir at stand 2 and alpine fir at stand 3 from 2009 to 2011. See text for further details. Discussion This study shows that emamectin is effective against C. occidentalis in three host tree species. Defoliation on treated trees was significantly lower than untreated trees over a 2-year period at stand 1 and a 3-year period at stands 2 and 3. There is some uncertainty in our conclusions because we did not quantify defoliation before treating the trees. A second possible limitation to our conclusions involved emamectin residue analysis over time. In other words, we cannot be sure if residual activity of emamectin was responsible for the differences between treated and control trees. However, we suspect that emamectin was responsible for the observed reduction in defoliation because defoliation on treated trees was always lower than control trees at different stands for the three tree species injected. Residue analysis would be beneficial to ascertain uptake and retention of emamectin over time and to calibrate dosage specifically to species of tree injected. For example, an understanding of the residual activity of emamectin over time may be necessary to refine treatment protocols depending on the host tree attacked by C. occidentalis, as has been suggested in studies of other insect pests of trees (Poland et al. 2006, Mota-Sanchez et al. 2009). Treatment of grand fir with emamectin benzoate met the target for suppression of eruptive forest Lepidoptera in Canada (SOPFIM 2012). However, the criterion was not met for Douglas-fir and alpine fir 2 and 3 years posttreatment. It is likely that the dosage and/ or formulation need to be adjusted depending on the species of host injected (residue analysis). Emamectin may also be used successfully against other pests of conifers in North America, such as the Douglas-fir tussock moth, Orgyia pseudotsugata McDunnough; western blackheaded budworm, Acleris gloverana (Walsingham); and spruce budworm, C. fumiferana Clemens on high-value trees, but a careful evaluation of efficacy is recommended. One injection of the TREE-äge solution may be insufficient for protection of foliage if heavy populations of C. occidentalis persist for 10 years at a location (Fellin and Dewey 1982, Campbell et al. 2006). Trees suffering infestations of this duration may require a second injection approximately 3–4 years after the first injection to insure protection of the live crown (Wright and Berryman 1978). Emamectin may also be used along with therapeutic applications of fertilizer or basal area reduction strategies to facilitate quick recovery from damage following the outbreak (Filip et al. 1992). Endnote 1. " Please see www.foresthealth.info/Flex/FPC. " We thank Pat Halseth, Rose Marie Helmer, and Gary Kempton for assistance with data collection, and David Cox (Syngenta) for reviewing an earlier version of this manuscript. Special thanks to the Idaho Panhandle National Forest for permission to install and evaluate treatments and USDA Forest Service, Forest Health Protection-Coeur d'Alene Field Office for assistance. " This article uses metric units; the applicable conversion factors are: centimeters (cm): 1 cm = 0.39 in.; meters (m): 1 m = 3.3 ft; kilometers (km): 1 km = 0.6 mi; hectares (ha): 1 ha = 2.47 ac. Literature Cited Arborjet, Inc . 2010 . TREE-age insecticide. Available online at shop.arborjet.com/ProductDetails.asp?ProductCode=0006 ; last accessed Feb. 26, 2013. Argentine J.A. Jansson R.K. Starner V.R. Halliday W.R. 2002 . Toxicities of emamectin benzoate homologues and photodegradates to Lepidoptera . J. Econ. Entomol. 95 ( 6 ): 1185 – 1189 . Google Scholar Crossref Search ADS PubMed WorldCat Campbell R. Smith D.J. Arsenault A. 2006 . Multicentury history of western spruce budworm outbreaks in interior Douglas-fir forests near Kamloops, British Columbia . Can. J. For. Res. 36 : 1758 – 1769 . Google Scholar Crossref Search ADS WorldCat Chukwudebe A.C. Andrew N. Drottar K. Swigert J. Wislocki P.G. 1996 . Bioaccumulation potential of 4“-epi-(methylamino)-4”-deoxyavermectin B1a benzoate (emamectin benzoate) in bluegill sunfish . J. Agric. Food Chem. 44 : 2894 – 2899 . Google Scholar Crossref Search ADS WorldCat Cowles R.S. Montgomery M.E. Cheah C.A.S.-J. 2006 . Activity and residues of imidacloprid applied to soil and tree trunks to control hemlock woolly adelgid (Hemiptera: Adelgidae) in forests . J. Econ. Entomol. 99 ( 4 ): 1258 – 1267 . Google Scholar Crossref Search ADS PubMed WorldCat Crookston N.L. 1985 . Forecasting growth and yield of budworm-infested Forests. II. Western North America and Summary . P. 214 – 230 in Recent advances in spruce budworms research: Proceedings of the CANUSA spruce budworms research symposium , Sanders C.J. Stark R.W. Mullins E.J. Murphy J. (eds). Ministry Sup. Serv. Canada , Ottawa, Ontario, Canada . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Fellin D.G. Dewey J.E. 1982 . Western spruce budworm . USDA For. Serv., For. Ins. Dis. Leaf. (FIDL) No. 53 . 10 p. OpenURL Placeholder Text WorldCat Fettes J.J. 1950 . The method for estimating defoliation . P. 89 – 90 in Investigations of sampling techniques for population studies of the spruce budworm on balsam fir in Ontario. For. Insect. Lab. Ann. Tech. Rep. Sault Ste. Marie , Ontario, Canada . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Filip G.M. Parks C.A. Hosman K.P. Wickman B.E. Mason R.R. 1992 . Thinning and nitrogen fertilization in a grand fir stand infested with western spruce budworm. III. Tree wound dynamics. For . Sci. 38 ( 2 ): 265 – 274 . OpenURL Placeholder Text WorldCat Fry H.R.C. Ryall K.L. Dixon P.L. Quiring D.T. 2008 . Suppression of Ennomos subsignaria (Lepidoperta: Geometridae) on Acer pseudoplatanus (Aceraceae) in an urban forest with bole-implanted acephate . J. Econ. Entomol. 101 ( 3 ): 822 – 828 . Google Scholar Crossref Search ADS PubMed WorldCat Furniss R.L. Carolin V.M. 1977 . Western forest insects . USDA For. Serv., Misc. Publ. No. 1339 , Washington, DC . 654 p. Google Scholar Crossref Search ADS Grosman D.M. Upton W.W. McCook F.A. Billings R.F. 2002 . Systemic insecticide injections for control of cone and seed insects in loblolly pine seed orchard–2 year results. S . J. Appl. For. 26 : 146 – 152 . OpenURL Placeholder Text WorldCat Grosman D.M. Clarke S.R. Upton W.W. 2009 . Efficacy of two systemic insecticides injected into loblolly pine for protection against southern pine bark beetles (Coleoptera: Curculionidae) . J. Econ. Entomol. 102 ( 3 ): 1062 – 1069 . Google Scholar Crossref Search ADS PubMed WorldCat Ioriatti C. Schmidt S. Pasqualini E. Civolani S. Anfora G. Angeli G. 2009 . Toxicity of emamectin benzoate to Cydia pomonella (L.) and Cydia molesta (Busck) (Lepidoptera: Tortricidae): Laboratory and field tests. Pest Manage . Sci. 65 ( 3 ): 306 – 312 . OpenURL Placeholder Text WorldCat Kline J.D. Azuma D.L. Alig R.J. 2004 . Population growth, urban expansion, and private forestry in western Oregon . For. Sci. 50 ( 1 ): 33 – 43 . OpenURL Placeholder Text WorldCat Kramer P.J. Kozlowski T.T. 1979 . Physiology of trees. Academic Press , New York . 642 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mason R.R. Wickman B.E. Paul H.G. 1997 . Radial growth response of Douglas-fir and grand fir to larval densities of the Douglas-fir tussock moth and the western spruce budworm . For. Sci. 43 : 194 – 205 . OpenURL Placeholder Text WorldCat Mota-Sanchez D. Cregg B.M. McCullough D.G. Poland T.M. Hollingworth R.M. 2009 . Distribution of trunk-injected 14C-imidacloprid in ash trees and effects on emerald ash borer (Coleoptera: Buprestidae) adults . Crop Prot. 28 : 655 – 661 . Google Scholar Crossref Search ADS WorldCat Mushtaq M. Allen L. R. S. Crouch L.S. Wislocki P.G. Fate of 3H- and 14C-labeled emamectin benzoate in lactating goats . J. Agric. Food Chem. 45 : 253 – 259 . Crossref Search ADS WorldCat O'Grodnick J.S. Wislocki P.G. Reynolds J.L. Wisocky M. Robinson R.A. 1998a . Aged soil column leaching of emamectin benzoate (MAB1a) . J. Agric. Food Chem. 46 : 2044 – 2048 . Google Scholar Crossref Search ADS WorldCat O'Grodnick J.S. Wislocki P.G. Keenan K.P. Beavers J.B. Frey L.T. Jaber M. 1998b . Subchronic and reproductive toxicity of emamectin benzoate to mallard ducks and northern bobwhite quail . Environ. Tox. Chem. 17 ( 11 ): 2318 – 2324 . Google Scholar Crossref Search ADS WorldCat Poland T.M. Haack R.A. Petrice T.R. Miller D.L. Bauer L.S. Gao R. 2006 . Field evaluations of systemic insecticides for control of the Asian longhorned beetle, Anoplophora glabripennis (Coleoptera: Cerambycidae), in China . J. Econ. Entomol. 99 ( 2 ): 383 – 392 . Google Scholar Crossref Search ADS PubMed WorldCat Potter D.A. Foss L. Baumler R.E. Held D.W. 2005 . Managing eastern tent caterpillars Malacosoma americanum (F.) on horse farms to reduce risk of mare reproductive loss syndrome. Pest Manage . Sci. 61 : 3 – 15 . OpenURL Placeholder Text WorldCat R Development Core Team . 2010 . R: A language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria . Available online at www.R-project.org ; last accessed Aug. 8, 2012. Sial A.A. Brunner J.F. 2010 . Toxicity and residual efficacy of chlorantraniliprole, spinetoram and emamectin benzoate to oblique banded leafroller (Lepidoptera: Tortricidae) . J. Econ. Entomol. 103 ( 4 ): 1277 – 1285 . Google Scholar Crossref Search ADS PubMed WorldCat Société de protection des forêts contre les insects et maladies (SOPFIM) . 2012 . The ABCs of spraying. Available online at www.sopfim.qc.ca/en/the-abcs-of-spraying.html ; last accessed May 6, 2013. Wright L.C. Berryman A.A. 1978 . Effect of defoliation by the Douglas-fir tussock moth on moisture stress in grand fir and subsequent attack by the fir engraver beetle (Coleoptera: Scolytidae) . USDA For. Serv., Res. Note PNW-323, Pacific Northwest Research Station , Portland, OR . 14 p. Copyright © 2013 by the Society of American Foresters.

York, Robert, A.;DeVries,, Rose

doi: 10.5849/wjaf.13-003pmid: N/A

Stem-snapping frequency and species differences were assessed in ∼30-year-old mid-elevation mixed-species plantations after a winter with above average, but not extreme, snowfall. Snapping (occurring between breast height and mid-crown) was not common (5/acre) but could be considered significant, depending on specific objectives and planning horizons. Black oak (Quercus kelloggii) snapped most often, especially compared with giant sequoia (Sequoiadendron giganteum), which did not snap at all. Ponderosa pine (Pinus ponderosa), white fir (Abies concolor), and Douglas-fir (Pseudotsuga menziesii) did not snap in proportions greatly different than prestorm densities. Snapped conifers tended to have greater height/diameter ratios than nonsnapped trees, but diameter ratios in these maturing plantations were generally low, probably a product of previous thinning treatments. Stem snaps could diminish or enhance objectives for modern mixed species plantations, suggesting the need for further study. stem snap, mid-elevation mixed conifer, height/diameter ratio, snow damage, snag development Forestland owners invest considerable resources into plantations for the long-term prospect of high value (Arney 2012). This investment is perpetually at risk, because factors such as pests and wildfire represent the potential for partial or complete loss. Less catastrophic, but nonetheless significant given the level of investment, is the loss of individual trees from snow damage during the several decades before final harvest. Managers may be able to reduce this risk, or at least hedge against it, by influencing species composition. Wonn and O'Hara (2001) found, for example, that larch trees in the Rocky Mountains were damaged less frequently than Pinus species. Such differences between species in mixed-conifer plantations of the Sierra Nevada, where multiple species are typically planted, could influence planting designs and thinning prescriptions. Alternatively, the total amount of damage might be so small as to be inconsequential. If damage is infrequent and occurs via snapping high enough along the stem to create usable snags for wildlife in an otherwise homogeneous structure, then storm damage may even be viewed as positive for improving wildlife habitat. We assessed differences in snapping (defined here as stems breaking between breast height and halfway up crowns) frequency after the winter of 2009–2010 in three mature plantations of the Sierra Nevada. While limited in time (1 year) and space (60 acres), species diversity within these particular plantations is high. Stem snapping was frequent enough to assess differences between ponderosa pine (Pinus ponderosa), giant sequoia (Sequoiadendron giganteum), Douglas-fir (Pseudotsuga menziesii), and white fir (Abies concolor). We also measured coppiced black oaks (Quercus kelloggii), which in these plantations are managed for wildlife food and habitat. The high tree diversity in these plantations reflects the trend toward managing for higher diversity in Sierra Nevada plantations. Information about the dynamics of the few mature multispecies plantations that exist is therefore in demand to inform young plantation management and to suggest areas of further study. To this end, we asked the following two questions: How common was stem snapping and were there differences in snapping frequency among species? Was height/diameter ratio (commonly used to predict damage) related to snapping probability? Methods Three plantations aged 29, 30, and 33 years were used. The plantations were within Blodgett Forest Research Station, between 4,000- and 4,300-ft elevation on the western slope of the Sierra Nevada range, California (38°52′ N; 120°40′ W). Forests in this region are productive and can be commercially thinned after around 30 years. After harvests, standard practices of site preparation, planting, herbicide, and precommercial thinning occurred. All the native conifer species (Douglas-fir, incense-cedar [Calocedrus decurrens], ponderosa pine, sugar pine [Pinus lambertiana], and white fir) were planted. Heavy mortality of incense-cedar and sugar pine precluded their inclusion in this study. Giant sequoia is not found locally but has been planted throughout the Sierra Nevada (Kitzmiller and Lunak 2012). Black oak were cut and allowed to sprout. During a preliminary analysis, we found no differences in snapping patterns among the plantations, which occur over a range of aspects (north, east, and southwest) and slope positions (from near ridge tops to near valley bottoms). Tree density primarily ranged from 100 to 200 trees/acre. These structures are representative of plantations in mid-elevation mixed-conifer forest. Annual precipitation averages 65 in., about one-third coming from winter snowfall (weather data come from a station at Blodgett Forest). The winter of 2009–2010 was not extreme, totaling 129 in. of snow compared with a 50-year average of 97 in. Stem snaps at this elevation tend to occur during and after storms. There were five storms that dropped more than 1 ft of snow within 24 hours (not an uncommon frequency). We measured snaps only if they occurred during the winter of 2009–2010 (as evident from green foliage retention). We used a network of plots on a 198 × 198 ft grid to sample the population of >7-in. dbh trees. A point-centered quarter sample selected the nearest trees to plot center in four surrounding quadrants. This proved to be an efficient method for sampling the infrequent population of snaps. Both the nearest live and snapped tree were measured (4 live and 4 snaps per plot). The search distance was limited to 100 ft to avoid trees being double-counted. Measuring distance from plots to trees allows for an estimate of density (Pollard 1971) of both snapped and live trees. Dbh and total height were measured for all trees (for snapped trees, height was measured directly by taping the portion on the ground). Analysis The density of both snapped and live trees was calculated for each plot with the equation where D is the density for a plot, n is the number of samples per plot, and r is the distance from the tree to the plot in the ith quadrant (convert to density by multiplying by 43,560 ft2/acre). Prestorm density was calculated by using the closest tree to plot in each quadrant, regardless of whether it was a snapped or live tree. Snapped trees were then expressed in terms of overall density and relative to prestorm density. Differences in snapping among species are assessed by calculating the proportions of sampled trees that were snapped for each species, as well as associated 95% confidence intervals. An initial logistic regression showed a significant (P < 0.01) interaction between species and height/diameter ratio in the prediction of snapping occurrence. Height/diameter ratio of all trees was then analyzed with a standard least squares model predicting height/diameter ratio from status (live or snapped) nested within species. This assessed the difference in height/diameter ratio between snapped and live trees for each species. Pairwise comparisons between snapped and nonsnapped trees were done for each species with Tukey's honestly significant difference tests. Values of P < 0.05 verified differences. Results and Discussion Stem snapping frequency was generally low, averaging less than 5/acre. Of all trees present before the storms, only 3% of them were damaged via snaps. A loss of 5 trees/acre repeated multiple times during the period leading up to rotation age, however, could cumulate to be significant. Whereas the snaps in this study occurred during a winter with above average snowfall, the individual storms and total amount of snow were not exceptional. Of the winters over the past 50 years at Blodgett Forest, 25% have had equal or greater amounts of snowfall. Even assuming a short rotation age of 50 years (Blodgett plans on 90 years), several more winters with snowfall of equal or greater magnitude are certain. It is uncertain, however, precisely how a highly variable future storm regime will interact with maturing plantations (Powers and Oliver 1970), given the potentially novel dynamics of mature plantations. Although our study domain is limited in terms of predictive power, it suggests a prudence of closely tracking storm damage in plantations to understand trends, factors, and impacts on timber value. Clearly, stem snapping affects values beyond timber. From the perspective of wildlife habitat, for example, a periodic input of snags could be beneficial. Snapped trees tended to be smaller (11 in. versus 14 in. dbh for live trees; t test P < 0.01), so utility for cavity nesting is limited. Snags were, however, created from the main canopy and will be created from larger trees in the future (they will also input more surface fuel and potential insect brood material). Although specific snag density targets are a rather narrow approach to managing snag habitat (Bagne et al. 2008), it is worth noting that the density of freshly created snags here is similar to targets on federal lands (USDA Forest Service 2003). The processes of individual tree mortality and growth differentiation, although often assumed to be extremely homogeneous, can create considerable structural diversity in mature mixed-species plantations. This process of diversification through snapping and growth differentiation is one that can be quantified as mixed-species plantations mature. The tendency of stems to snap varied considerably by species (Table 1), resulting in a change in relative species composition (Figure 1). Although black oak was a relatively minor component (3% of stem density), it snapped with higher frequency compared with prestorm density (18% of all snaps were black oak) and therefore became even more rare. Black oak was the most vulnerable to snapping; giant sequoia was by far the most resistant. There were no sequoia snaps, compared with a prestorm relative density of 10%. After noting this during fieldwork, we did a complete grid search and still found no stem snapping. Other species did not snap in proportions greatly different from prestorm density. Ponderosa pine was slightly higher, whereas white fir and Douglas-fir were slightly lower. Table 1. Number of trees measured and the relative proportion of trees that had snapped stems by species in mixed-conifer plantations of the central Sierra Nevada. CI, confidence interval; NA, not applicable. Open in new tab Table 1. Number of trees measured and the relative proportion of trees that had snapped stems by species in mixed-conifer plantations of the central Sierra Nevada. CI, confidence interval; NA, not applicable. Open in new tab Figure 1. Open in new tabDownload slide Storm-caused stem snapping differences among species in Blodgett Forest, California. Positive values reflect a tendency to snap in proportions greater than prestorm density, and negative values indicate a lower probability. Figure 1. Open in new tabDownload slide Storm-caused stem snapping differences among species in Blodgett Forest, California. Positive values reflect a tendency to snap in proportions greater than prestorm density, and negative values indicate a lower probability. From a timber perspective, the loss of black oak could be considered beneficial, because there is currently no local market for black oak sawlogs. These plantations were designed to retain black oak, however, so the loss of black oak is concerning. Black oak is arguably “worth more alive than dead,” as it produces acorn masts and can eventually create unique habitat structures in older plantations. The comparatively sturdy giant sequoia was unambiguously positive from a timber perspective. It outgrew all other conifers by a considerable margin (82 ft versus 62 ft) and will probably maintain dominance in individual tree size as long as thinning occurs periodically (York et al. 2013). The common perceptions of sturdy oaks and brittle giant sequoia are ill-deserved, at least in these maturing plantations. Differences in black oak and giant sequoia morphology suggest possible explanations. Black oak was relatively short in stature (45 ft versus the stand-wide average of 62 ft), so competitive position was low. Because of weak epinastic control (Oliver and Larson 1996), the crowns of black oak have large branches that grow at snow-retaining angles, and crowns can bend severely from plagiotropic responses to light (Kramer and Kozlowski 1979). Finally, resistance to cold temperature while burdened with snow may be particularly low for black oak. The tradeoff between increased hydraulic conductive capacity and decreased protection from cavitation (embolism) has long been identified as a basic difference between conifers and hardwoods (Cochard and Tyree 1990). In contrast, giant sequoia's tall and narrow crowns with many small branches (York et al. 2013) could be particularly efficient at shedding snow. Finally, giant sequoia's stem durability was observed directly during later commercial thins in these stands, as no logs were broken even after encouraging operators to “be rough” with them (R.A. York, pers. observ., May 20, 2012). Snapped ponderosa pine had greater height/diameter ratios (59:1) compared with nonsnapped trees (54:1), a small difference that was detectable because of large sample size (greater ratios indicate more “spindly” forms). A ratio of 80:1 was a damage threshold found for ponderosa pine in the northern Rocky Mountains (Wonn and O'Hara 2001), but nearly all of the ponderosa pine trees in this study, whether snapped or not, had ratios of less than 80:1. By comparison, snapped white fir had greater ratios (81:1) than nonsnapped trees (60:1), as did Douglas-fir (83:1 versus 60:1 for nonsnapped trees). The 80:1 threshold may apply to white fir and Douglas-fir in these plantations, although more samples would be needed to test this. Giant sequoia, which did not snap, had a relatively low ratio of 48:1. Because diameter growth is more sensitive to competition than height growth (Oliver and Larson 1996), maintaining low densities decreases height/diameter ratios and enhances individual tree resistance to storm damage (Wonn and O'Hara 2001). These plantations were precommercially thinned to approximately 170 trees/acre, a density much lower than what is typically found on nonmanaged lands experiencing fire suppression. Had stem density been maintained at higher levels, snapping probably would have been more common. There was no detectable difference (P = 0.26) in height/diameter ratio for black oak. Interestingly, however, the height/diameter ratio for black oak was greater for live trees (74:1) than for snapped trees (62:1). Although being tall and skinny makes conifers more susceptible to snapping, this may not hold for black oak. Managing plantations for the long-term persistence of black oak is an interesting challenge requiring further study. One possible tool for recruiting black oak is thinning sprout clumps, which may lead to a faster growth rate of individuals and greater competitive stature later on. Another is managing the density of surrounding competition. Maintaining a low density of conifers surrounding oaks may provide greater resource availability and improve individual vigor, but it may increase vulnerability to storm damage by causing horizontal growth and lateral branches. Further studies altering levels of both sprout thinning and neighbor thinning in mixed-species plantations will be of particular interest as modern plantations are managed for a variety of objectives in the future (Paquette and Messier 2009). " This project was supported by the Sponsored Projects for Undergraduate Research at UC Berkeley's College of Natural Resources. We thank Frieder Schurr for supplying climate data and Ken Somers, who reviewed an earlier version of this article. Literature Cited Arney J.D. 2012 . Working forest costs, 2012 PNW Reforestation Council presentation. Available online at www.forestbiometrics.com/presentations/2012-pnw-reforestation-council-presentation/attachment/working-forest-costs/ ; last accessed Jan. 29, 2013. Bagne K.E. Purcell K.L. Rotenberry J.T. 2008 . Prescribed fire, snag population dynamics, and avian nest site selection . For. Ecol. Manage. 255 : 99 – 105 . Google Scholar Crossref Search ADS WorldCat Cochard H. Tyree M.T. 1990 . Xylem dysfunction in Quercus: Vessel sizes, tyloses, cavitation and seasonal changes in embolism . Tree Physiol. 6 : 393 – 408 . Google Scholar Crossref Search ADS PubMed WorldCat Kitzmiller J.H. Lunak G. 2012 . Growth of giant sequoia compared to ponderosa pine and other mixed conifers in California plantations. West . J. Appl. For. 27 ( 4 ): 196 – 204 . OpenURL Placeholder Text WorldCat Kramer P.J. Kozlowski T.T. 1979 . Physiology of woody plants. Academic Press , New York . 811 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Oliver C.D. Larson B.C. 1996 . Forest stand dynamics. John Wiley and Sons , New York . 520 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Paquette A. Messier C. 2009 . The role of plantations in managing the world's forests in the Anthropocene . Front Ecol Environ. 8 : 27 – 34 . Google Scholar Crossref Search ADS WorldCat Pollard J.H. 1971 . On distance estimators of density in randomly distributed forests . Biometrics 27 : 991 – 1002 . Google Scholar Crossref Search ADS WorldCat Powers R.F. Oliver W.W. 1970 . Snow breakage in a pole-sized ponderosa pine plantation. USDA For. Serv., Res. Note PSW-218, Pacific Southwest Forest and Range Experiment Station , Los Angeles, CA . 3 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC USDA Forest Service . 2003 . Sierra Nevada forest plan amendment. Pacific Southwest Region, R5-MB-019 . 382 p. Wonn H.T. O'Hara K.L. 2001 . Height:diameter ratios and stability relationships for four northern Rocky Mountain tree species . West. J. Appl. For. 16 ( 2 ): 87 – 94 . OpenURL Placeholder Text WorldCat York R.A. O'Hara K.L. Battles J.J. 2013 . Density effects on giant sequoia (Sequoiadendron giganteum) growth through 22 years: Implications for restoration and plantation management . West. J. Appl. For. 28 ( 1 ): 30 – 36 . Google Scholar Crossref Search ADS WorldCat Copyright © 2013 by the Society of American Foresters.