https://orcid.org/0000-0002-9346-5194
Abstract
The Cutting Methods Study at the Ford Forest in the Upper Peninsula of Michigan, USA, was established in 1956 and has been maintained continuously on a 10-year cycle. Methods consist of three diameter limits (DL; 13, 30, and 41 cm), single-tree selection to three residual basal area limits (STS; 11, 16, and 21 m2ha−1), and light improvement (LI) focused on improving tree grade. Long-term results show that the 41 cm DL produced the greatest managed forest value and cumulative sawlog production, followed by the STS to 11 m2ha−1 residual basal area. STS treatments and LI were uniformly superior at improving standing tree grade. In contrast, treatments that emphasize removal of large diameter trees while retaining moderate residual basal area (the 41 cm DL and 11 m2ha−1 STS) produced the largest harvest volumes of high-grade sawlogs, driving financial performance. Stand density has declined in all treatments except the 30 and 41 cm DL, where it has increased, and these two treatments have larger abundance of saplings and poles. Alternative partial cutting methods such as selection to lower residual basal areas and medium-intensity diameter-limit cuts thus may provide greater financial returns and higher average quality, and could have implications on regeneration and long-term sustainability.
Long-term comparison of alternative partial cutting practices in northern hardwoods in the Upper Peninsula of Michigan over 60 years reveals that Arbogast-based single-tree selection is inferior using financial and volume yield criteria. Alternatives that remove more of the larger trees appear over time to increase regeneration and harvested tree quality, which in turn drives financial performance. However, treatments with extremely high volume removals perform poorly against all others, and have few, if any, redeeming financial, silvicultural, or ecological qualities.
Single-tree selection is a silvicultural system that is widely prescribed in Lake States northern hardwoods and has an established history of use throughout the entire range of the forest type (Kern et al. 2014, Pond et al. 2014). At the start of the 20th century, partial cutting methods were well established in Europe, and questions had begun to arise about their use in the United States (O’Hara 2002, Kenefic and Kern 2013). To help answer these questions, studies testing the viability of several different partial cutting methods in northern hardwoods were instituted in the Lake States (Eyre and Zillgitt 1953, Reed et al. 1986, Niese et al. 1995). And indeed, initial research in the Lake States showed that single-tree selection in uneven-aged northern hardwoods could shape and maintain an economically and ecologically productive forest (Eyre and Zillgitt 1953, Erdmann and Oberg 1973). The management goals of stands under single-tree selection usually include high-quality sawtimber and a balanced diameter distribution (Nyland 1998, Tubbs 1977a). To help achieve these goals, analysis of stand structure and financial returns are useful evaluations that can provide important information about quality and value.
Quality and volume directly affect financial gain and consequently also influence management decisions. Previous long-term studies evaluating silvicultural treatments in northern hardwoods have found that single-tree selection can increase quality and volume growth (Johnson 1984, Leak and Sendak 2002). The monetary value of most hardwoods is greatly affected by wood quality, and so managing to improve quality and reduce logging damage in northern hardwoods is important for improving financial yield; with an increase in quality and grade, the monetary value of a tree sharply increases (Miller 1991, Wiedenbeck and Smith 2018). Increases in quality resulting from single-tree selection in hardwoods have resulted in higher financial yield (Orr et al. 1994, Schuler et al. 2017). In a simulation in northern hardwoods, Bohn et al. (2011) found that initial harvest volumes will be greatest with diameter-limit cuts but may be later outperformed by single-tree selection with up to 20%–40% greater volumes per harvest. A simulation by Nyland (2005) also found that selection treatments will generate higher long-term revenues. However, in several long-term studies, diameter-limit cuts resulted in the greater financial returns than selection (Reed et al. 1986, Smith and Miller 1987, Erickson et al. 1990, Miller 1993, Buongiorno et al. 2000). A recent study in northern hardwoods also found that single-tree selection cuts did not greatly increase growth or survival in residual trees (Moreau 2019). This past research shows the value of long-term studies to test hypotheses based on simulation and also exhibits that there can be successful growth under various different partial cutting management regimes. Thus, there is a need to evaluate the long-term success and applicability of different management regimes over many harvest entries, using both financial and ecological factors (Erickson et al. 1990).
Data from the Cutting Methods Study at the Ford Forest present an exceptional opportunity to evaluate long-term consequences from silvicultural alternatives in northern hardwoods. The study was established in 1956, with an objective of evaluating stand development and financial benefit under different partial cutting methods through time (Bourdo 1957). Methods tested include single-tree selection, cutting above a diameter limit, a “light improvement” cut emphasizing residual quality over structure, and an unmanaged control. Although the design uses single replicates of each cutting method, this case study and associated long-term dataset have generated results that increase the evidence base on which solutions to management problems can be formed. An important precedent can be found in the historic northern hardwoods Cutting Methods Study at the Dukes Experimental Forest in the Upper Peninsula of Michigan. Although also unreplicated, it is nevertheless the foundation for the renowned Eyre and Zillgit (1953) publication, on which the influential and widely prescribed Arbogast (1957) marking guide for northern hardwoods is directly based. The Cutting Methods Study at the Ford Forest was designed in consultation with Carl Arbogast with the intent to establish selection treatments similar to those at the Dukes Experimental Forest (Kern et al. 2014). Unfortunately, there are problems with the Dukes dataset, including gaps in harvest cycle entries and damage to the control, making other long-term studies, such as this one, that have been consistently maintained and monitored in upper Great Lakes northern hardwoods forest, even more consequential.
Two important publications have resulted from the Cutting Methods Study at the Ford Forest: Reed et al. (1986) and Erickson et al. (1990), which present 22 and 32 year results. In addition to examining yield and quality improvement, these studies contrast financial performance of the various treatments in terms of discounted net revenues to the initiation of the study (in 1956) and in terms of a long-term sequence of repeatable yields once stand structure has stabilized under several harvest cycles sensuRideout (1985). These alternative approaches embody two perspectives, one of the landowner considering near-term financial implications of alternatives and the other a long-term perspective on the continuous yield and financial return. Given that a defining feature of selection silviculture is continuous stocking, when assessing the long-term consequences of alternative methods, the return under repeated cycles of the same treatment is arguably much more relevant than return from initial conversion.
To evaluate long-term performance, both Reed et al. (1986) and Erickson et al. (1990) assumed that stand structure and periodic yield had stabilized 20–30 years after the initial conversion. A key conclusion from both analyses was the superiority of diameter-limit methods over single-tree selection, based on financial criteria, particularly for a diameter limit of 41 cm, in contrast to larger or smaller alternatives. These early studies also suggest that careful application of the selection system led to improvements in the grade of residual standing trees, in comparison with both diameter-limit treatments and an untreated control. Although some shifts in ranking among the treatments occurred in the 10-year interval between the two analyses, both questioned whether increases in grade under the selection treatment would eventually lead to an increase in return that would surpass those of the diameter-limit cuts, a result forecast by Bohn et al. (2011) and Nyland (2005) using models. These questions can only be addressed empirically following an extended study period, and as Reed et al. (1986) note, they are “of utmost importance to small, nonindustrial ownerships.”
The overall goal of this analysis was to reexamine this silvicultural experiment (Reed et al. 1986, Erikson et al. 1990), with an emphasis on changes that have accumulated over the second 30 years of the study. Specific objectives included: (1) to examine stability of yield and structure within and across treatments since the most recent analysis, (2) to explore the hypothesis that further improvement in standing tree grade would occur in single-tree selection but not diameter-limit treatments, (3) to revisit the hypothesis that single-tree selection treatments could “catch up” to the diameter-limit treatments on financial criteria, and (4) to identify factors that may explain observed differences in yield, structure, and financial performance with time, which may be explored in future research.
Methods
Site Description
The study is located at the Ford Forest in Baraga County, Michigan (46.66° N, 88.51° W), which is a research forest owned and operated by Michigan Technological University. The 26.3 ha study area is uniform in soils and topography, and as a consequence, is also uniform in ecological habitat type and site productivity. Soils are classified as Alouez gravelly coarse sandy loams, slopes 0%–6%, habitat type ATD (Acer-Tsuga-Dryopteris; Burger and Kotar 2003), and sugar maple site index is 19–21 m base age 50 (Soil Conservation Service 1988). Circa 1898, the white pine in the area was removed from what had initially been a pine-hardwood forest, and subsequent selective logging in the early 1900s removed over 70% of the merchantable sawtimber (Bourdo 1957, Reed et al. 1986). In 1956, the Cutting Methods Study was begun on the residual northern hardwoods forest, which was an uneven-aged sugar maple-dominated mix of regeneration and remnant trees from previous harvests. When age data were collected in 2014, the overstory sugar maple in the control ranged from 39 to 253 years old, and after 52 years of management, the age structure of all six treatments was similar (Previant 2015).
The study area was divided into eight different silvicultural treatments and a control, ranging from 1.2 to 5.7 ha in area, with an objective of showing the effects of various cutting methods on stand development and yield. Treatments are: (1) 13 cm diameter-limit (DL13), (2) 30 cm diameter-limit (DL30), (3) 41 cm diameter-limit (DL41), (4) 56 cm diameter-limit (DL56), (5) single-tree selection system cut to 21 m2ha−1 residual basal area (STS21), (6) single-tree selection system cut to 16 m2ha−1 residual basal area (16 STS), (7) single-tree selection system cut to 11 m2ha−1 residual basal area (STS11), (8) light improvement (LI), and (9) control (Bourdo 1957; see supplemental data for detail). The LI treatment is defined as a 15–16 m2ha−1 residual basal area treatment based on increasing quality. In the selection treatments, a balanced diameter distribution with a q-value of 1.3 and a maximum tree size of 61 cm were goals for each harvest entry. Selection treatment guidelines followed Arbogast (1957) and emphasized removing trees that had poor quality or vigor, were high risk or cull, lacked the potential for improvement, or impeded the growth of other high-quality trees. In diameter-limit cuts, harvesting was confined to trees above the diameter limit with no tending in the residual stand.
From 1956 to the present day, the treatments have been harvested on 10-year cycles, though some were cut only when volume growth had been sufficient, in the judgement of the supervising forester, to make an operational-scale harvest feasible. Efforts were consistently made throughout the study period to reduce site damage and potential harm to residual trees, although growth and survival of individual trees is primarily related to the competitive preharvest environment when compared with potential machinery damage (Moreau 2019). Skid trails present within compartments were reused and crews were usually local, familiar with the area, and in regular contact with the supervising forester.
Measurements
Within each treatment unit, a single 0.4 ha plot was established and divided into 10 equal-area subplots, with a preharvest inventory conducted in 1956. The following measurements were collected for trees greater than 12.6 cm diameter at breast height (dbh): species, dbh, merchantable sawlog height to a 25.4 cm outside bark minimum diameter, gross and net volume (international quarter-inch rule) for trees greater than 28 cm dbh, cull percent, and butt-log tree grade. The same measurements were also collected prior to harvests in 1957, 1968, 1978, 1988, and 1998. Prior to harvest in 2008, slightly different measurements were collected. These included, for trees greater than 12.6 cm dbh: species, dbh, location within plot, total height, height to live, merchantable height, and percent soundness and grade for sawtimber trees. Prior to harvest in 2018, measurements collected were species, dbh, number of 2.4 m sawlogs and 2.4 m pulp sticks, and percent soundness and grade for sawtimber trees. Standing tree grades were estimated following the USDA Forest Service Timber Management Field Book Region 9 (USDA Forest Service 2008) and volumes were estimated using Gevorkiantz and Olsen (1955).
At each harvest entry, the harvest removal from the entire treatment unit was stacked and scaled by grade onsite. Logs were graded to the standards of the Northern Hardwood and Pine Manufacturers Association prior to 1988, and after, to the standards of the Timber Producers Association of Michigan and Wisconsin. These data were standardized to a per-unit-area basis for financial analyses, and volume was converted from board feet to cubic meters following Winn et al. (2020).
Growth and Yield
Estimated butt-log grade for sawtimber size class trees was used to indicate stand quality improvement over time. Postharvest standing volumes and cumulative volume removals by grade were used to compare outcomes from the different treatments. The effect of treatment on stand structure was evaluated by comparing prestudy diameter distributions from 1956, preharvest diameter distributions from halfway through the study in 1998, and the most recent preharvest diameter distributions from 2018 (Pond and Froese 2015). Unfortunately, preharvest stand structure data from 1988 were not available, so 1998 data were used instead.
Species abundance was used to evaluate changes in overstory (trees >12.6 cm dbh) composition since study establishment. Pretreatment 1956 values were taken directly from the establishment report (Bourdo 1957) and preharvest 2018 values were found using individual tree data. All values are reported as a percentage of stems per unit land area relative to each species, for each treatment.
Financial Analysis
In this study, managed forest value (MFV) was emphasized in financial analyses, following Rideout (1985). MFV is the discounted value of the sum of an infinite series of harvests, assuming a stable structure and reliable estimate of future volume removals by grade, and is intended to represent the long-term applicability of a silvicultural system. Sustainable stand structure following Rideout (1985) presupposes that all future harvests will produce similar cash flow each cutting cycle. In this study, two assumptions about sustainable stand structure were evaluated: (1) that it was reached in 1978, after two entries, following Reed et al. (1986) and Erickson et al. (1990), and (2) in 1998, relying instead on the most recent 30 years of results since Erickson (1990). Financial performance was compared between each treatment and the control using discount rates of 2% and 4%. Total revenue accumulated since 1978 was also evaluated at these same discount rates.
In both MFV and total revenue calculations, 2018 dollars were used to facilitate comparison. Whereas MFV for 1998–2018 uses current 2018 prices, the price structure used in total revenue calculations was that of regional historic prices from each year of harvest adjusted to a common 2018 basis using the Producer Price Index for Lumber and Wood Products (US Bureau of Labor Statistics 2020). Actual prices were used where available from timber sale records, augmented where necessary by historic pricing from Ulrich (1983) and Michigan Department of Natural Resources stumpage reports (Michigan Department of Natural Resources 2000). The MFV values for 1978–1988, taken directly from Erickson et al. (1990), were also adjusted to a 2018 basis. When comparing treatment performance, costs may not be included when the end result is a decision whether or not to take management action (Martin and Leudeke 1989); therefore, no cost factors were introduced into the analysis, due to the intent to compare relative, not absolute, measures of financial returns between treatments.
Results
Quality
The grade distribution for preharvest standing tree volume continually improved from 1968 through 2018 in the LI and the STS treatments, with the most notable change being a decline in the proportion of nonmerchantable and sharp increase in grade 1 (Figure 1). The greatest increase in percentage of grade 1 trees since 1968 of the selection treatments was in the STS11 (increase by 283%) and by 2018 the proportion of grade 1 had increased from about 15%–20% to 30%–50% in all STS treatments. In contrast, although the grade distribution improved somewhat between 1968 and 1988 in the DL treatments, there has been little change through 2018, and the proportion of grade 1 actually fell in all but the DL56 treatment. In 2018, the STS11 had the greatest proportion of grade 1 among the STS treatments, and the DL41 among the DL treatments. Still, over the entire duration of the study, grade 1 percentages in all treatments have increased, excluding the DL30 and DL13.
Preharvest standing tree butt log grade distribution (percent) by grade as measured in 1968, 1988, and 2018 in 0.4-hectare measurement units, for trees > 28 cm dbh (Erickson et al. 1990). Below grade class consists of trees >28 cm dbh without any merchantable volume, as well as cull trees. Grade 1 includes veneer.
Notably, by 2018, only the LI and STS11 had greater grade 1 percentages than the control. Because grade is strongly determined by tree size, the grade increase in the control with time likely reflects an increase in mean tree size, and thus comparisons may be confounded with the absence of treatment. To isolate treatment differences, grade proportion was recalculated for only trees ≥46 cm dbh; this dbh was chosen because, under the log grading rules used, all trees 46 cm and above had equally the least stringent requirements to meet grade 1 criteria (smallest minimum length per clear cutting and greatest maximum number of clear cuts per face). Under this criterion, in 2018, all STS treatments and the LI had a larger proportion grade 1 standing tree volume than the control (see supplemental data). For trees ≥30 cm dbh and ≤46 cm dbh, that is, trees likely to increase in grade as diameter increases, the DL41 had the highest number of grade 2 trees both pre- and post-harvest.
Volume and Yield
Standing sawtimber volumes through time have persisted as expected, with the DL56, DL41, and DL30 treatments each retaining more volume than the next smaller limit, respectively (Figure 2). Similarly, the residual volumes of the selection treatments line up as STS21 > STS16 > STS11. Since 1988, the DL41 treatment consistently had the greatest sawtimber volume growth, with an average of 2.92 m3ha−1 per year (Figure 2). Prior to harvest in 2018, all treatments except for the harshest diameter limits (DL30 and DL13) had greater per hectare volumes than in 1956.
![Net residual standing sawtimber volume (m3ha−1, from 0.4 ha measurement units) over the 62-year study period [international quarter-inch rule, following Gevorkiantz and Olsen (1955)]. Conversion from board feet to cubic meters following Winn et al. (2020). Postharvest values not available for 1968 and 1978 harvests. A random number drawn from the range (−0.5, 0.5) was added to year to reduce overplotting.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/forestscience/67/5/10.1093_forsci_fxab024/1/m_fxab024f0002.jpeg?Expires=1690968787&Signature=FPkpLJ4cA2Ky1fcCqZypj5I7c-xDjkzBLW9UuyuWk72mjFHgJsfiNM6fmkwvXgHH3R-EWrxm~ZG6Br8Y1pvcJI4FHTdMAkE8LzAMJQVL7Z2khK2Qe6OWXkpQmjbshRAxwuOkHEu0OrdMCpKBvjQMtte5uVm7lNxJLkmMFzOdycPCwUA3iIaTk1vDXlsBjVsnduut3dEkNsqd1duTFuDzjcO1wM6iui~z1AWiBn7Q2ikyBWJzL3mQxFJufJezay0oSlQuK1JyKZGSOHF2T55PNpd8dVJcNZhf3xu3VIKuggx0Vg6-zDhtJU-suvLSAbIvjPbCiMolEfGvIxsL3D9wfg__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Net residual standing sawtimber volume (m3ha−1, from 0.4 ha measurement units) over the 62-year study period [international quarter-inch rule, following Gevorkiantz and Olsen (1955)]. Conversion from board feet to cubic meters following Winn et al. (2020). Postharvest values not available for 1968 and 1978 harvests. A random number drawn from the range (−0.5, 0.5) was added to year to reduce overplotting.
Sawtimber volume yields since 1978 for the selection treatments follow the expected pattern of volume from STS11 > STS16 > STS21. The STS21 was not harvested in 2018, although standing volume was arguably sufficient for an operational harvest, and so the potential yield may be underestimated at least for that entry. Among diameter-limit treatments, the DL41 produced greater yields than the DL30 or DL56 (Figure 3). The DL41 treatment had not only the greatest overall removal volume, with an average of 2.2 m3ha−1 per year, but the greatest total volumes of grade 1 and grade 2 sawtimber harvested since sustainable stand structure was reached in 1978. The grades for these harvested volumes are more accurate than those estimated for standing trees.
![Net scaled harvested sawtimber volume by grade (percent of total m3ha−1) and treatment [international quarter-inch rule, following Gevorkiantz and Olsen (1955)] from 1978–2018. Volumes derived from entire-treatment unit removals. Conversion from board feet to cubic meters following Winn et al. (2020). Grade 1 includes veneer.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/forestscience/67/5/10.1093_forsci_fxab024/1/m_fxab024f0003.jpeg?Expires=1690968787&Signature=yvWB~YmG3BxjBLmX4Zwrk1B2EEjp89Dr2-HvSeex85xoFXPrv8~pMDhiSrDE1LRPoNGFVUibPPldyVpDNftFrUYPKKeopj~yoEcAIgbAZwNLWS4UeIJYvTShOxMnYVRLnw6HxHceiUZrMsf1FVpEwGvcX2j7ZZuPlCG3-jomPE3cCr7lp23SoPHVn7KRds5-TURm4E8ifsGYwE7VO5-Fb54n742Evpw7L48fE9VWbMvgxV8HCmlhT1YR5-OTu8NQW9UBXMUgmpudX-RuoMUJ7x-wmv-abn49DhUfrMRoJ1OFmWMyuANXhQxPz2X9zK7NsmtzQuoKO4Dklv1g~~29wQ__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Net scaled harvested sawtimber volume by grade (percent of total m3ha−1) and treatment [international quarter-inch rule, following Gevorkiantz and Olsen (1955)] from 1978–2018. Volumes derived from entire-treatment unit removals. Conversion from board feet to cubic meters following Winn et al. (2020). Grade 1 includes veneer.
Financial Returns
Excluding 1957–1968, and thus the effects of initial stand conditions, the DL41, STS21, and LI performed best under MFV at a 4% discount rate in the first half of the study (Erickson et al. 1990; Table 1). Using only returns from the previous three decades, the DL41 still ranked first, but was followed instead by the STS11 and STS16 (Table 1). Real sugar maple stumpage prices were higher in 2018 than 1988 (Wagner et al. 2019), resulting in greater dollar values for the 1998–2018 MFV and periodic return calculations.
Comparison of managed forest value (MFV) and periodic return from the first two decades after sustainable stand structure was assumed, and from the subsequent three decades, with sawlog products valued at price in 2018 dollars. Discount rates of 2% and 4% are shown.
| Period | Attribute | DL56 | DL41 | DL30 | DL13 | STS21 | STS16 | STS11 | LI |
|---|---|---|---|---|---|---|---|---|---|
| 1978–1988a | Periodic return ($·ha−1) | 133 | 566 | 1060 | 1060 | 261 | 191 | 250 | 525 |
| Mean treatment interval (years)b | 10 | 10 | 30 | 40 | 10 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 609 | 2584 | 1306 | 878 | 1190 | 872 | 1142 | 2397 | |
| MFV ($·ha−1) 4% | 277 | 1177 | 472 | 279 | 1095 | 398 | 521 | 1093 | |
| Ranking | 8 | 1 | 5 | 7 | 2 | 6 | 4 | 3 | |
| 1998–2018 | Periodic return ($·ha−1) | 2051 | 2240 | 1159 | 1425 | 1880 | 1781 | 2005 | 1145 |
| Mean treatment interval (years)b | 17 | 10 | 20 | 40 | 13 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 5126 | 10 228 | 2385 | 1180 | 6403 | 8134 | 9155 | 5227 | |
| MFV ($·ha−1) 4% | 2164 | 4664 | 973 | 375 | 2827 | 3709 | 4174 | 2384 | |
| Ranking | 6 | 1 | 7 | 8 | 4 | 3 | 2 | 5 | |
| ∆ MFV rank | +2 | 0 | −2 | −1 | −2 | +3 | +2 | −2 |
| Period | Attribute | DL56 | DL41 | DL30 | DL13 | STS21 | STS16 | STS11 | LI |
|---|---|---|---|---|---|---|---|---|---|
| 1978–1988a | Periodic return ($·ha−1) | 133 | 566 | 1060 | 1060 | 261 | 191 | 250 | 525 |
| Mean treatment interval (years)b | 10 | 10 | 30 | 40 | 10 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 609 | 2584 | 1306 | 878 | 1190 | 872 | 1142 | 2397 | |
| MFV ($·ha−1) 4% | 277 | 1177 | 472 | 279 | 1095 | 398 | 521 | 1093 | |
| Ranking | 8 | 1 | 5 | 7 | 2 | 6 | 4 | 3 | |
| 1998–2018 | Periodic return ($·ha−1) | 2051 | 2240 | 1159 | 1425 | 1880 | 1781 | 2005 | 1145 |
| Mean treatment interval (years)b | 17 | 10 | 20 | 40 | 13 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 5126 | 10 228 | 2385 | 1180 | 6403 | 8134 | 9155 | 5227 | |
| MFV ($·ha−1) 4% | 2164 | 4664 | 973 | 375 | 2827 | 3709 | 4174 | 2384 | |
| Ranking | 6 | 1 | 7 | 8 | 4 | 3 | 2 | 5 | |
| ∆ MFV rank | +2 | 0 | −2 | −1 | −2 | +3 | +2 | −2 |
a Values from Erickson et al. 1990 adjusted to 2018 prices using the Producer Price Index for Lumber and Wood Products. Real sugar maple stumpage prices were higher in 2018 than in 1988, contributing to differences in value.
b Treatment interval is the average of the cutting cycles lengths associated with each harvest by the end of the period, except for the 13DL where it is the interval between the only two harvests in the entire study.
Comparison of managed forest value (MFV) and periodic return from the first two decades after sustainable stand structure was assumed, and from the subsequent three decades, with sawlog products valued at price in 2018 dollars. Discount rates of 2% and 4% are shown.
| Period | Attribute | DL56 | DL41 | DL30 | DL13 | STS21 | STS16 | STS11 | LI |
|---|---|---|---|---|---|---|---|---|---|
| 1978–1988a | Periodic return ($·ha−1) | 133 | 566 | 1060 | 1060 | 261 | 191 | 250 | 525 |
| Mean treatment interval (years)b | 10 | 10 | 30 | 40 | 10 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 609 | 2584 | 1306 | 878 | 1190 | 872 | 1142 | 2397 | |
| MFV ($·ha−1) 4% | 277 | 1177 | 472 | 279 | 1095 | 398 | 521 | 1093 | |
| Ranking | 8 | 1 | 5 | 7 | 2 | 6 | 4 | 3 | |
| 1998–2018 | Periodic return ($·ha−1) | 2051 | 2240 | 1159 | 1425 | 1880 | 1781 | 2005 | 1145 |
| Mean treatment interval (years)b | 17 | 10 | 20 | 40 | 13 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 5126 | 10 228 | 2385 | 1180 | 6403 | 8134 | 9155 | 5227 | |
| MFV ($·ha−1) 4% | 2164 | 4664 | 973 | 375 | 2827 | 3709 | 4174 | 2384 | |
| Ranking | 6 | 1 | 7 | 8 | 4 | 3 | 2 | 5 | |
| ∆ MFV rank | +2 | 0 | −2 | −1 | −2 | +3 | +2 | −2 |
| Period | Attribute | DL56 | DL41 | DL30 | DL13 | STS21 | STS16 | STS11 | LI |
|---|---|---|---|---|---|---|---|---|---|
| 1978–1988a | Periodic return ($·ha−1) | 133 | 566 | 1060 | 1060 | 261 | 191 | 250 | 525 |
| Mean treatment interval (years)b | 10 | 10 | 30 | 40 | 10 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 609 | 2584 | 1306 | 878 | 1190 | 872 | 1142 | 2397 | |
| MFV ($·ha−1) 4% | 277 | 1177 | 472 | 279 | 1095 | 398 | 521 | 1093 | |
| Ranking | 8 | 1 | 5 | 7 | 2 | 6 | 4 | 3 | |
| 1998–2018 | Periodic return ($·ha−1) | 2051 | 2240 | 1159 | 1425 | 1880 | 1781 | 2005 | 1145 |
| Mean treatment interval (years)b | 17 | 10 | 20 | 40 | 13 | 10 | 10 | 10 | |
| MFV ($·ha−1) 2% | 5126 | 10 228 | 2385 | 1180 | 6403 | 8134 | 9155 | 5227 | |
| MFV ($·ha−1) 4% | 2164 | 4664 | 973 | 375 | 2827 | 3709 | 4174 | 2384 | |
| Ranking | 6 | 1 | 7 | 8 | 4 | 3 | 2 | 5 | |
| ∆ MFV rank | +2 | 0 | −2 | −1 | −2 | +3 | +2 | −2 |
a Values from Erickson et al. 1990 adjusted to 2018 prices using the Producer Price Index for Lumber and Wood Products. Real sugar maple stumpage prices were higher in 2018 than in 1988, contributing to differences in value.
b Treatment interval is the average of the cutting cycles lengths associated with each harvest by the end of the period, except for the 13DL where it is the interval between the only two harvests in the entire study.
Total discounted revenues since 1978, when sustainable stand structure was assumed to have been reached (Reed et al. 1986), show the rate at which different treatments have been accumulating revenue (Figure 4). At all discount rates, the DL41 was uniformly superior. Other treatments, such as the STS11 and LI that have been cut each 10-year harvest cycle also perform well, but still did not show a rate of revenue accumulation that has allowed them to equal or surpass the DL41. The STS11 and STS21 began in 1978 with very similar revenue, but the STS11 rapidly surpassed the STS21 at all discount rates. The selection treatments, DL41, and DL30 have all accumulated revenue at a greater rate than the DL56 and DL13 over the past three decades.
Cumulative revenue since 1978 in 2018 dollars by treatment: (A) 0% discount rate, (B) 2% discount rate, and (C) 4% discount rate.
Structure and Density
When this study began in 1956, the diameter distribution followed a reverse-J shape that was relatively homogenous across treatments (i.e., in pretreatment condition; Figure 5A). By 1998, differences in preharvest maximum diameter were clearly apparent among the DL treatments and the control but less so among the STS treatments (Figure 5B). Between 1998 and 2018, structure was relatively stable in the DL treatments, with a sharp drop in density at the relevant diameter limit. In contrast, structure in the STS treatments appears more variable within than between treatments, but appears to have become more stable and more closely approximating the original reverse-J shape in 2018 (Figure 5B, C). Preharvest, in both 1998 and 2018, the DL30 and DL41 had the most trees in the pole and small sawtimber size classes, with the disparity between them and all other treatments growing larger with time. The 1998 and 2018 preharvest structures of the selection treatments can still be described as reverse-J, but stocking in the smallest diameter classes was much lower than that measured in 1956.
(A) Stand structure pre-harvest at study commencement in 1956 for trees ≥12.6 cm dbh. The 50 cm class contains all trees in larger diameter classes. (B) Stand structure pre-harvest in 1998 for trees ≥12.6 cm dbh. (C) Stand structure pre-harvest in 2018 for trees ≥12.6 cm dbh.
The DL30, DL41, and DL56 treatments have similar postharvest basal areas to the STS11, STS16, and STS21 treatments, respectively (Figure 6B). However, there is a clear difference in trees per hectare (TPH) between selection and diameter-limit treatments in that the diameter-limit treatments are associated with increasing stand density with time. For the past four harvest entries, these three diameter-limit treatments had much higher postharvest TPH than the single-tree selection treatments; even the DL56 treatment outperforms the selection treatments here. So TPH has increased over time for the diameter-limit treatments while steadily decreasing for the selection treatments and control (Figure 6A). This indicates the diameter-limit treatments are recruiting new trees, net of mortality and removals, whereas the selection treatments are not. The noticeable decline in postharvest TPH without a corresponding decline in basal area (ideally kept at 11, 16, 21, and 15–16 m2ha−1 for the STS11, STS16, STS21, and LI treatments respectively) also indicates that growth in selection treatments is increasingly accumulating on larger diameter trees.
Stand density over time for all treatments, excluding the DL13, in terms of (A) number of trees (stems·ha−1) and (B) basal area (m2·ha−1). Postharvest values not available for 1968 and 1978 harvests. A random number drawn from the range (−0.5, 0.5) was added to year to reduce overplotting.
Species Composition
Relative overstory species composition pretreatment in 1956 was dominated by sugar maple, with the STS21 having the lowest relative sugar maple abundance at 72% and the LI having the highest at 94% (Table 2). Sugar maple abundance increased in all three single-tree selection treatments from 1956 to 2018, as well as in the control, DL30, and DL56, with the STS21 and DL56 having the greatest gains (+13%) in sugar maple relative abundance. Notably, the DL41 had the greatest decrease in relative abundance of sugar maple (−7%) from 1956 to 2018. In 2018, the control had the highest relative abundance of sugar maple at 97%. Yellow birch relative abundance decreased with time in all treatments except for the STS11.
Initial 1956 pretreatment relative species abundance (percent) and total change to preharvest 2018, for trees >12.6 cm dbh, in 1956 and 2018.
| Treatment | Acer saccharum | Betula alleghaniensis | Ulmus americana | Othera | ||||
|---|---|---|---|---|---|---|---|---|
| Initial | ∆ | Initial | ∆ | Initial | ∆ | Initial | ∆ | |
| DL56 | 76 | +13 | 6 | −2 | 6 | −5 | 11 | −6 |
| DL41 | 86 | −7 | 7 | −6 | 6 | +1 | 1 | +12 |
| DL30 | 79 | +1 | 6 | −6 | 6 | −3 | 8 | +9 |
| DL13 | 88 | -4 | 3 | −3 | 5 | +2 | 4 | +5 |
| STS21 | 72 | +13 | 12 | −9 | 12 | −11 | 4 | +7 |
| STS16 | 80 | +4 | 14 | −11 | 3 | 0 | 3 | +7 |
| STS11 | 76 | +10 | 8 | 0 | 8 | −2 | 8 | −8 |
| LI | 94 | −4 | 0 | 0 | 6 | +1 | 0 | +3 |
| Control | 90 | +7 | 5 | −4 | 1 | −1 | 4 | −2 |
| Treatment | Acer saccharum | Betula alleghaniensis | Ulmus americana | Othera | ||||
|---|---|---|---|---|---|---|---|---|
| Initial | ∆ | Initial | ∆ | Initial | ∆ | Initial | ∆ | |
| DL56 | 76 | +13 | 6 | −2 | 6 | −5 | 11 | −6 |
| DL41 | 86 | −7 | 7 | −6 | 6 | +1 | 1 | +12 |
| DL30 | 79 | +1 | 6 | −6 | 6 | −3 | 8 | +9 |
| DL13 | 88 | -4 | 3 | −3 | 5 | +2 | 4 | +5 |
| STS21 | 72 | +13 | 12 | −9 | 12 | −11 | 4 | +7 |
| STS16 | 80 | +4 | 14 | −11 | 3 | 0 | 3 | +7 |
| STS11 | 76 | +10 | 8 | 0 | 8 | −2 | 8 | −8 |
| LI | 94 | −4 | 0 | 0 | 6 | +1 | 0 | +3 |
| Control | 90 | +7 | 5 | −4 | 1 | −1 | 4 | −2 |
a Other may include one or more of: Betula papyrifera, Picea glauca, Prunus serotina, Ostrya virginiana, Tilia americana, or Tsuga canadensis.
Initial 1956 pretreatment relative species abundance (percent) and total change to preharvest 2018, for trees >12.6 cm dbh, in 1956 and 2018.
| Treatment | Acer saccharum | Betula alleghaniensis | Ulmus americana | Othera | ||||
|---|---|---|---|---|---|---|---|---|
| Initial | ∆ | Initial | ∆ | Initial | ∆ | Initial | ∆ | |
| DL56 | 76 | +13 | 6 | −2 | 6 | −5 | 11 | −6 |
| DL41 | 86 | −7 | 7 | −6 | 6 | +1 | 1 | +12 |
| DL30 | 79 | +1 | 6 | −6 | 6 | −3 | 8 | +9 |
| DL13 | 88 | -4 | 3 | −3 | 5 | +2 | 4 | +5 |
| STS21 | 72 | +13 | 12 | −9 | 12 | −11 | 4 | +7 |
| STS16 | 80 | +4 | 14 | −11 | 3 | 0 | 3 | +7 |
| STS11 | 76 | +10 | 8 | 0 | 8 | −2 | 8 | −8 |
| LI | 94 | −4 | 0 | 0 | 6 | +1 | 0 | +3 |
| Control | 90 | +7 | 5 | −4 | 1 | −1 | 4 | −2 |
| Treatment | Acer saccharum | Betula alleghaniensis | Ulmus americana | Othera | ||||
|---|---|---|---|---|---|---|---|---|
| Initial | ∆ | Initial | ∆ | Initial | ∆ | Initial | ∆ | |
| DL56 | 76 | +13 | 6 | −2 | 6 | −5 | 11 | −6 |
| DL41 | 86 | −7 | 7 | −6 | 6 | +1 | 1 | +12 |
| DL30 | 79 | +1 | 6 | −6 | 6 | −3 | 8 | +9 |
| DL13 | 88 | -4 | 3 | −3 | 5 | +2 | 4 | +5 |
| STS21 | 72 | +13 | 12 | −9 | 12 | −11 | 4 | +7 |
| STS16 | 80 | +4 | 14 | −11 | 3 | 0 | 3 | +7 |
| STS11 | 76 | +10 | 8 | 0 | 8 | −2 | 8 | −8 |
| LI | 94 | −4 | 0 | 0 | 6 | +1 | 0 | +3 |
| Control | 90 | +7 | 5 | −4 | 1 | −1 | 4 | −2 |
a Other may include one or more of: Betula papyrifera, Picea glauca, Prunus serotina, Ostrya virginiana, Tilia americana, or Tsuga canadensis.
Discussion
The continuation of the Cutting Methods Study through 2018 extends the results reported at year 22 (by Reed et al. 1986) and year 32 (by Erickson et al. 1990) to a total of 62 years. A key finding after the additional 30 years is the consistent superiority of the DL41 treatment over all others on both financial, harvested grade, and volume production criteria. This is a surprising result, given diameter-limit cutting is routinely critiqued as either inferior to selection because of the lack of tending in residual size classes to improve tree quality (e.g., Niese et al. 1995, Kenefic and Nyland 2005, Kenefic et al. 2005, Fajvan 2006), or ultimately dysgenic by the repeated removal of trees of superior vigor that grow beyond the diameter limit and are removed in future cycles (e.g., Nyland 1988, Howe 1989, Erickson et al. 1990, Buongiorno et al. 2000, O’Hara 2002). Indeed, in the Cutting Methods Study, no tending was performed in the residual trees below the diameter limit across the entire 62-year duration, and over the past 30 years, there has been no improvement in residual tree grade. Paradoxically, however, the DL41 has consistently and reliably produced the highest average annual yield of any treatment, greatest grade 1 harvested volumes, and a superior distribution of harvested tree grade that is exceeded only recently by the LI and STS21 (Figure 3).
The first 32 years of the study demonstrated clear improvements in residual tree grade under all of the treatments, including in the reserve. Average standing tree grade can increase in time because inferior trees are targeted for removal, but also, trees can simply grow large enough to cross size thresholds that delimit higher grades. By examining the grade of only trees ≥46 cm dbh, it appears to be clear that grade improvements have occurred through management. The key unanswered question posed by Reed et al. (1986) and Erickson et al. (1990) was whether, in time, one or more of the STS treatments could “catch up” to the DL41 because of an increasing grade profile. The results here clearly demonstrate a steady increase in standing tree grade in the STS and in particular in the LI since 1988, along with an associated increase in harvested tree grade. This makes sense, as eventually, if inferior trees are removed, only the superior trees remain to be harvested. However, these increases have not translated into increases in revenue that, even without discounting, have proved superior to the DL41 (Figure 4).
The failure of the STS or LI treatments to financially surpass the DL41 is, in retrospect, not particularly surprising. Financial return depends on both grade profile of harvested trees and periodic yield. Given that the threshold diameter for a grade 1 sawlog is 41 cm (USDA Forest Service 2008) in the DL41, soon after trees cross the threshold, they are removed, leaving behind only smaller grade 2 and 3 trees that may increase in grade through the next cycle as they grow beyond size thresholds. Because STS treatments will retain some trees that already meet the size criteria for grade 1, much of the increase in value in those trees can come only from volume growth; however, growth is penalized by discounting, and these treatments in general have lower periodic yield than the DL41. Moreover, by maintaining some stocking in trees as large as 61 cm, any grade 2 trees that are approaching the 41 cm threshold are deprived of growing space, leading to slower growth. The same logic can explain the inferiority of the DL56 and DL30 to the DL41. With a large threshold diameter, the DL56 carries many grade 1 trees to the next cycle and overall high residual basal area seems equated with low periodic yield. In contrast, although the DL30 has had high periodic yield over the life of the study, the low threshold diameter ensures that trees rarely, if ever, grow large enough to meet the much more valuable grade 1 specification before they are harvested in the next entry.
The DL41 is effectively a “sweet spot”, at least in this maple-dominated forest type, reentry cycle, and level of site productivity, holding the greatest number of smaller sawtimber-sized grade 2 trees that are poised to increase significantly in grade and value between cycles (Godman and Mendel 1978, Smith et al. 1979, Reed and Mroz 1997, Webster et al. 2009, Power and Havreljuk 2018). Because the DL30 also holds a large proportion of grade 2 trees postharvest, it too has the potential for significant grade improvement. A lower diameter threshold for grade 1, a longer reentry cycle, or higher site productivity would likely improve the financial performance of the DL30. The STS11 also holds a significant proportion of grade 2 trees (Figure 1) and, combined with comparatively high periodic growth, has shown strong financial performance, especially in the last 30 years. Ultimately, however, whereas the STS11 is perhaps the closest of the STS treatments to the “sweet spot” occupied by the DL41, the retention of some large trees beyond the grade 1 diameter threshold means some opportunity cost will always occur. This cost is only amplified by the potential loss of high-value residual trees due to destructive weather or degrade due to damage or decay. For example, in the Lake States, there is an issue with increasing dark heart as trees age, and this defect degrades logs that would otherwise meet grade 1 or higher standards (Erickson et al. 1992). To minimize this opportunity cost, Reed and Mroz (1997) suggested favoring removal of larger grade 1 and 3 trees if single-tree selection is the chosen method. This captures value in grade 1 trees before any future degrade and concentrates growth on grade 2 trees that still contain some quality sawlogs.
Although the top-performing treatment by MFV has been consistently the DL41, significant shifts in ranking have occurred since 1988. For example, Erickson et al. (1990) suggested that the LI receive special attention because of significant improvement in residual tree grade over the first 32 years and excellent performance based on MFV. However, although the LI was ranked 2nd in 1988, it has since been surpassed by the STS11 and STS16, which are now ranked 2nd and 3rd, respectively (Table 1). By 2018 the LI easily has the best standing tree and harvested log grade profile, but middling and declining periodic yield do not allow the superior grade profile to be translated into outstanding financial return. In contrast, whereas the improvement in harvested grade was more modest in the STS11 and STS16, periodic yield has increased substantially. Since 1988, the already poorly ranked DL30 and DL13 treatments have further decreased in ranking, suggesting that extremely low residual basal area can be problematic. The DL13 treatment in particular has no redeeming qualities; it has extremely low MFV, poor structural diversity, and is easily outperformed in average sawtimber volumes harvested per year over the last 40 years by every other treatment. Harsh diameter-limit cuts like the DL13 and DL30 also may present greater potential to accumulate dysgenic effects through time (Buongiorno et al. 2000).
Divergence of structure and density between the selection and diameter-limit treatments was present by postharvest 1988, with the diameter-limit treatments retaining more trees in the pole and smaller sawtimber size classes (Figure 5) and increasing in TPH where selection treatment stem numbers have consistently decreased (Figure 6). However, structure since then has appeared to stabilize, with the same differences present in 2018. Consistency in structure and density lends greater weight to the financial analysis results due to the implication of sustained future yield; density is particularly important because it does affect quality in hardwoods, as higher density causes better form in survivor trees (Godman and Brooks 1971, Sonderman 1985). The greater numbers of stems present in the DL41 over the past 30 years bode well for the sustainability of its current quality and volume yield into the future.
Other investigations have found that medium-intensity diameter-limit cuttings can produce sustained volume yield. Smith and Miller (1987) reported that a 40 cm diameter-limit cut had the greatest periodic annual volume production (log scaled) when compared with two selection treatments and a clearcut, as well as the greatest compounded periodic harvest value. Results from Buongiorno et al. (2000) also suggest that sustainable management of Lake States northern hardwoods with diameter-limit cutting is possible; a 38 cm cut performed similarly to a heavy selection cut, with high overstory diversity as well as high present value and periodic harvest values. Interestingly, both Smith and Miller (1987) and Buongiorno et al. (2000) recommend that a medium-intensity diameter-limit cut could be improved by removing poorly performing understory trees to increase future quality. This would effectively result in a diameter-limit treatment with stand improvement and would likely increase value and performance of treatments like the DL41 even more, a conclusion that can be tested in future studies or through actual implementation by managers in the field.
As currently recommended in the Lake States, single-tree selection retains residual stocking levels of 16 to 21 m2ha−1 and is concerned with obtaining a specific residual reverse-J diameter distribution (Eyre and Zillgitt 1953, Arbogast 1957, Gilbert and Jensen 1958, Tubbs 1977a, Nyland 1998), a structure commonly found in balanced, uneven-aged stands and widely accepted as a representation of such (Leak et al. 1987, Leak 1996, Nyland 2016). However, long-term application has not only been associated with a lack of regeneration but also with a reduction in overstory species and structural diversity (Angers et al. 2005, Neuendorff 2007, Gronewold et al. 2010). Silvicultural methods that vary harvest intensity and stand structure can help mitigate or avoid the decline in overstory diversity that accompanies traditional single-tree selection (Hanson and Lorimer 2007), and in some studies, partial cutting methods including diameter-limit cuts and selection to lower residual basal areas have been associated with higher overstory species diversity (Niese and Strong 1992, Buongiorno et al. 2000). Indeed, in this study, the DL41 had the greatest decrease in sugar maple relative abundance since 1956, and abundance increased in all three selection treatments. Selection treatments that cut to lower residual basal areas than traditionally recommended have also been shown to help accommodate regeneration of less shade tolerant species (Tubbs 1977b, Leak et al. 1987, Bodine 2000), and the STS11 was the only treatment to not decrease in relative abundance of the intermediately tolerant Betula alleghaniensis Britt. (Table 2). Heavier cuttings and disturbance in northern hardwoods long-term management have been found to result in a rotated sigmoid diameter distribution that could be accompanied by an increase in softwoods and species diversity (Leak 1996). Higher intensity partial cutting also helps prevent the homogenization of understory stand structure common with single-tree selection (Angers et al. 2005), while increasing overstory species diversity (Niese and Strong 1992), both factors important to wildlife habitat. Lower residual basal areas such as those found in the DL41, DL30, and STS11 also result in higher light availability, a factor associated with greater regeneration densities in Great Lakes northern hardwoods (Matonis et al. 2011). This is shown in how the DL41 and DL30 are doing a better job than the selection treatments of continually recruiting more trees into the smaller size classes (Figure 5), contributing to structural diversity and implying greater levels of regeneration. Diameter-limit cuts have also been found to have a higher genetic diversity and contain rare alleles that may in the future be adaptively beneficial, which is particularly relevant in the context of climate change (Hawley et al. 2005).
Overall, the DL41 and STS11 treatments perform well financially while also providing motivations apart from financial gain to consider alternatives to traditional single-tree selection in northern hardwoods, due to their favorable performance under ecological metrics such as species and structural diversity. This is especially important given concerns about the lack of regeneration under traditional uneven-aged single-tree selection silviculture (Previant 2015, Bassil et al. 2019, Hupperts et al. 2019). Due in part to those concerns, evaluating alternatives has become a point of interest for forest managers, with recent studies calling for alternatives to selection silviculture as it is currently practiced in northern hardwoods (Matonis et al. 2011, Hupperts et al. 2019, Henry et al. 2021) while also finding that it is frequently applied in a manner inconsistent with accepted standards (Pond et al. 2014). Recognizing that alternatives to traditional single tree-selection, such as the DL41 and STS11, have the potential to address concerns about increased species diversity, structural diversity, and regeneration in northern hardwoods is increasingly significant in the face of climate change and prospective future changes in disturbance regimes (D’Amato et al. 2011, Campione et al. 2012, Gustafson et al. 2020).
Supplementary Material
Supplementary material is available at Forest Science online.
Supplement 1. Detailed description of treatments including harvest entry years since study establishment.
Supplement 2. Grade distribution preharvest and postharvest in 2018, by size classes corresponding to northern hardwoods log grading guidelines.
Acknowledgments
We are grateful for the vision and investment made by the faculty, staff, and students at the Ford Center and Forest in maintaining the Cutting Methods Study for over 60 years. Special appreciation is given to Jim Schmierer for preharvest data collection and management of the 2018 harvest entry.
Funding
This work was supported by a grant from the Michigan Sustainable Forestry Initiative Implementation Committee and the College of Forest Resources and Environmental Science at Michigan Technological University.
