2. Materials and Methods
The specific Montréal Process C&I included in this paper (
Table 1) were selected based on data availability and applicability towards cross-ownership comparisons. All data sources needed to be citable, be able to be subset by ownership category, and cover most, if not all, of the conterminous USA. Preference was given to quantitative data and sources that allowed for temporal comparisons. The analysis was constrained to the conterminous USA due to data limitations for Alaska and Hawaii in the national forest inventory and other data sources.
The forest ownership categories used (
Table 2) were based on a collapsing of the ownership classes used by the US Department of Agriculture (USDA), Forest Service, Forest Inventory and Analysis program [
11], which was the primary data source used to quantify C&I in this analysis (
Table 1). Under the broader grouping of Private forest ownership, the forestland was classified as Corporate, Family, and Other Private. The Public ownership category was subdivided into Federal, Local, and State. And the final grouping was Tribal, which was not further subdivided; in previous reports (e.g., Renewable Resource Planning Act assessments [
12]), Tribal lands were typically grouped with other ownership categories, but these lands are part of sovereign nations, are neither Private nor Public, and hence we opted to classify this ownership group separately.
For half of the C&I analyzed (including some of Criterion 1 and all of Criteria 2, 3, and 5), the primary data source was the USDA Forest Service, Forest Inventory and Analysis plot data (FIA) [
13], which constitutes the national forest inventory for the USA. This is the same data source that was used as a primary data source in previous national assessments of forest sustainability in the USA [
14,
15]. The FIA program includes over 300,000 permanent inventory plots across the USA that span all ownership categories and on which a consistent set of forest mensuration and related data are collected. Forestland as defined by FIA, and, as used in this analysis, includes land at least 0.4 ha (1 ac) in size and 37 m (120 ft) wide with at least 10 percent tree canopy cover or has had at least 10 percent tree canopy cover in the past and is likely to have it again in the future (e.g., areas where clearcutting has occurred, but the forest is likely to regenerate through natural or artificial means) [
11]. Stand size was defined, following the FIA protocol, based on stocking level and the size of the plurality of trees on a plot. FIA defines stand sizes as: large for stands with ≥50% stocking in medium and large diameter trees and stocking of large diameter trees greater than or equal to stocking of medium diameter trees; medium for stands with ≥50% stocking in medium and large diameter trees and stocking of large diameter trees less than stocking of medium diameter trees; small for stands with ≥50% stocking in small diameter trees; and non-stocked for stands with an all live stocking values of <10 (base 100) [
11]. Large diameter trees are defined as trees ≥28 cm diameter at breast height (DBH) for hardwoods and ≥23 cm DBH for softwoods; medium as trees ≥13 cm DBH and <28 cm for hardwoods and ≥13 cm DBH and <23 cm for softwoods; and small as trees <13 cm DBH. Productivity was assessed by estimating the forest area that was capable of producing at least 1.4 m
3 ha
−1 yr
−1 (20 ft
3 ac
−1 yr
−1), an FIA threshold for productivity [
11]. Volume related estimates were for trees at least 13 cm (5 in) in diameter at breast height, also an FIA defined threshold. The area-weighted, mean remeasurement intervals were 7.1 years (nominal years of 2012 and 2019) for forest area and volume-based variables and 5.5 years variables (nominal years of 2013 and 2019) for growth, removals and land use change, the difference largely being due to a lack of data availability for the Rocky Mountain region (see
Supplemental Materials for specific inventory cycles used).
The Protected Areas Database (PAD) [
16] was used to estimate area of forestland by protection status (C&I 1.1.b) by ownership group. Data were summarized based on Gap Analysis Project (GAP) codes, which categorized lands as: 1 = permanently protected and mandated to retain a natural state; 2 = permanently protected and mandated to maintain a primarily natural state; 3 = permanently protected, but extractive uses permitted; 4 = not permanently protected. In addition to these GAP statuses, a separate variable indicated if the land was permanently protected via a conservation easement. FIA inventory plots were overlaid on the PAD layer and based on ownership at plot center were summarized as, in order of precedence, GAP 1, GAP 2, GAP 3, Easement, or Unprotected.
Forest area density was used as a proxy for forest fragmentation (C&I 1.1.c) [
7]. The proportion of the 15.2 ha area around each FIA inventory plot that was forested was calculated based on data from the US National Land Cover Database (NLCD) [
17]. Potential values ranged from 0 to 1 with higher values representing more forestland in the vicinity and implying lower fragmentation. The most recent estimates were based on 2016 NLCD data and the change data compared 2006 to 2016.
At-risk species (C&I 1.2b) were assessed using NatureServe’s Map of Biodiversity, which summarizes the number of threatened and endangered species listed under the U.S. Endangered Species Act and other imperiled or critically imperiled species identified by NatureServe within 990 m x 990 m grid cells [
18]. The FIA inventory plots were overlaid on this layer and summarized based on forest ownership at plot center. As the map of biodiversity values were simple counts, there was no way to filter for just at-risk forest species.
Economic benefits, i.e., jobs and wages, generated by the forests (C&I 6.3.a and 6.3.b) were estimated using data from the Bureau of Labor Statistics including the logging, sawmill, and pulp mill sectors and the support activities for forestry sector [
19]. The contributions by ownership category for forest sector employment and inflation-adjusted wages were estimated based on the relative contributions to harvest removals for the North, South, and Pacific Coast regions; in the Rocky Mountain region, where harvesting data were not available, the harvesting distribution was estimated from the share of public and private removals [
12] and further refined by shares of productive forestland [
20].
For Criteria 4 and 7, national-level, quantifiable data that could be disaggregated by ownership category were not available. Instead, literature syntheses were conducted where available literature on the respective indicators were summarized. For Criterion 4, the search terms “forestry”, “best management practices”, or the common abbreviation of this term (“BMP”), and “implementation” were combined with the name of each state as well as with “United States” using the search engine DuckDuckGo to find relevant reports and articles at the state and national level; specifically, we looked for resources that compared BMP implementation rate by ownership type. Other resources were snowball-sampled from the original reports and articles. For Criterion 7, resources were snowball sampled beginning with resources identified using the search terms “sustainable forest management”, “United States”, and “policy” or “legislation” for 7.1.a. and “clarity” or “security”, “land tenure”, and “United States” for Indicator 7.3.a. Previous summaries of Criteria 7 for forestland in the USA [
21,
22,
23] and information on Tribal land from the Indian Forest Management Assessment Team [
24] and heirs’ property were also specifically included.
Where necessary, data were subset by region (
Figure 1). These regions were based on those used by the USDA Forest Service’s Renewable Resource Planning Act Assessment [
12].
Changes were calculated for indicators where estimates were available for two points in time. For C&I 1.1.a, 2.b, and 2.d, absolute changes in terms of percentages were calculated as:
where
xi,t was the value for variable
x for ownership category
i at time
t. To avoid confounding trends with overall forest area trends, relative changes reported in terms of changes in percentage points were calculated for C&I 2.a, 2.c, 3.a, and 3.b:
where
Ai,t was the forest area for ownership category
i at time
t and other variables were defined as above. For C&I 1.1.c and 1.2.a, changes reported were simple differences (i.e.,
).
All data analyses were performed in the R statistical environment [
25].
4. Discussion
There are many C&I that have similar patterns across ownership categories, but there are also important differences. A large difference that has implications for many other attributes is the area covered by each ownership group and where these holdings are distributed across the USA (
Figure 2). For example, due to historical factors, such as colonization/homesteading and public acquisition patterns, most of the Federal forestlands are concentrated in the western USA and much (but certainly not all) of this land is of lower productivity. In addition to the total forestland areas, the juxtaposition of the ownerships can be equally important as most forestland exists within a matrix of different ownership types that collectively influence the broader landscape characteristics and must be considered when developing landscape-level, cross-boundary initiatives [
41].
Another important difference across ownership categories is the trends in forest areas (
Table 3C). While the area of all Public and Tribal ownerships did not vary by more than 1% (and all had gains), Family and Other Private forestlands decreased by 7.7% and 1.4%, respectively, and Corporate forestland increased by 6.9%. Of particular concern is the loss of forestland to non-forest uses, including 627,000 ha yr
−1 and 33,000 ha yr
−1 from Family and Other Private forestlands, respectively (
Table 3C). But of the land that is deforested, it is important to note that roughly half of the conversion is to agriculture and pasture [
42], which is more likely to revert to forest cover as compared to other land uses, such as development.
The net loss of forestland from an ownership category does not mean that all of that acreage was converted to non-forest uses. Much of the loss of Family and Other Private forestlands is transfers to other ownership categories with Corporate being the largest sink [
43]. Part of this transition is due to changes in ownership structure that are conducted primarily for tax purposes (e.g., a family or individual converting their holdings into a limited liability corporation) with few, if any, changes in management intent, and FIA’s inability to track these changes, but part is also due to investment and forestry companies acquiring Family forestland. Without a thorough analysis of the management practices and outcomes among these different ownership groups, it is impossible to determine whether this “loss” within ownership categories represents a problem or an opportunity from a sustainability perspective.
The net changes of forest area at the ownership category level fail to capture some important trends within each category. For example for Corporate forestland, there has been a dramatic shift in the last few decades from vertically-integrated forest products companies to institutional investors, including Timber Investment Management Organizations (TIMOs) that buy and sell forestland on behalf of institutional investors and individuals of high net worth, and Real Estate Investment Trusts (REITs), which are publicly traded companies that focus on real estate investments [
44]. Among Family forest ownerships, the average land tenure is fairly long (21 years [
1]), but these lands will eventually be sold or otherwise transferred to the next owners and this will result in opportunities for parcellation or consolidation of holdings [
45]. All of these trends have potential implications in terms of forest retention, forest management practices, and, more broadly, forest sustainability.
Forest area status and trends set the base for all other C&I because trends in forest area can confound interpretation of other C&I. All of the totals, such as growing stock volume, are directly impacted by forest area trends. Examining relative statistics (e.g.,
Table 3C) can be useful, but this is only a partial solution. For example, the relative area of planted forests within an ownership category can be calculated, but a trend such as the reduction in percentage of area of Corporate forestland that has been planted (
Table 4C) is not due to a net decrease in total area of planted forests; the absolute area actually increased, but the decrease was due to an influx of forestland from other ownership categories that was not planted.
The contributions of forests to biodiversity, at least in terms of number of tree species, varies considerably depending on whether absolute or relative statistics are examined. While Family forestland has the greatest number of tallied tree species, in part due to this ownership category representing a plurality of the nation’s forestland, Other Private and State forestlands have substantially higher relative numbers of trees species, 55 and 43 species per million ha, respectively, compared to other ownership categories. This is a reflection of these ownership categories being concentrated in areas of higher (tree) diversity and, to a certain extent, the ownerships focusing on acquiring rarer and more diverse properties.
For at-risk species, the greatest total area is again for Family forestland, but in relative terms all Private ownership categories are at similar levels, ranging from 62 to 65% of the forest area having one or more at-risk species, and the Public and Tribal categories vary substantially from 47% for Federal to 84% for State. Again, this is likely related to the location of these lands—with marked differences in the distribution of ownerships across ecoregions varying substantially in their baseline biodiversity. Although there was no way to separate at-risk forest species from other at-risk species, it is important to note that FIA plots with as little as 10% forest cover can be designated “forest,” and the parcels containing plots can have significant areas of non-forest cover types. For these reasons, although forest-associated at-risk species are probabilistically more likely to be found on FIA forest plots, no species can be categorically excluded. As such, the total number of at-risk species remains an appropriate index of the maximum potential number of endangered and at-risk species associated with a given FIA plot.
In terms of forest management, specifically timber harvesting (C&I 2.d), the patterns differ substantially across ownership categories. The plurality of the annual wood harvested comes from Corporate forestland followed closely by Family forestland, and the relative harvesting intensities, 3.0 and 1.4 m
3 ha
−1 yr
−1, respectively (
Table 4B), reflect this trend. This is logical given the importance of financial returns to many Corporate [
44] and some Family [
1] ownerships and the constraints and relatively lower emphasis placed on this activity for other ownerships. The lower harvesting rates are one of the factors influencing the higher per hectare volumes for Federal ownerships. The ratios of net growth to removals are above 1.0 for the totals and 90% of the most voluminous species. The ownership/species combinations with ratios of less than 1.0 are largely a result of pests, fires, and other disturbances negatively impacting growth. There are other species that may not be nationally prevalent but are important at the regional or local level that also have net growth to removal ratios of less than 1.0, but these too often face other threats, such as
Fraxinus americana (American white ash) being decimated by
Agrilus planipennis (emerald ash borer). Although the ratio of net growth-to-removals provides a convenient metric of the sustainability of species or groups of species across a managed landscape, there is significant potential for this metric to overestimate the margins available for sustained yield, especially where operational constraints or ownership objectives can lead to significant acreage remaining effectively reserved [
46]. Under these conditions, localized pockets of unsustainable loss within a given ownership category may persist within a region characterized by overall ratios greater than 1.0 for one or more species.
Damage is occurring across all ownership categories, but the relative proportions are highest for Federal forestland (
Table 3B), which due to historical reasons are more likely to be in arid and fire dependent ecosystems as compared to other ownership categories. The specific vectors, i.e., specific diseases, pests, or types of weather events, vary by geography. This partially explains the differences observed, but there are also mitigating factors due to management, or lack of management, occurring on some lands.
Trends in standing carbon and carbon sequestration are directly related to volume trends. Both carbon and volume trends are influenced by the cumulative impacts of harvesting, disturbance, productivity, and other factors. For example, the large net increases on Other Private and Family forestlands are a result of the relatively high growth rates compared to harvesting and other disturbances. The net decrease on Federal forestlands is largely driven by losses to disturbances, such as pests and wildfire.
Forest area density, a proxy of forest fragmentation (C&I 1.1.c), is declining across all ownership classes from 2006 to 2016 implying increased fragmentation. Tribal ownerships show a substantially lower value for this metric across the study period as compared to other ownership groups. This is likely due to a concentration of Tribal forestland in the arid west, rather than differences in management or land use transitions. Across all other ownerships, density declined by an absolute change of 0.05 (from an average 0.74 to 0.69) across the remeasurement period.
One especially important difference across ownership categories is related to if and how ownerships can influence the causes of the observed trends, which is partially captured by Criterion 7. The Legal and Institutional Frameworks assessed by this criterion describe the broad constraints within which the owners and managers operate and have enormous consequences on forest sustainability. For example, most Public forestland will stay undeveloped for perpetuity. Private forestland, apart from that with conservation easements, have a higher, although still relatively low, probability of conversion, but Private lands also have a higher probability of being a source of new forests through, for example, afforestation of marginal agricultural land. The high percentage of Public, and in particular Federal, forestland that is protected is due to the legislative mandates that established and dictate the uses of these lands. Conservation easements are the only mechanism for permanently conserving Private forestlands with 6% of these lands under an easement according to the Protected Areas Database [
16], although there are likely more easements on private lands that are missing from this database due to no complete national repositories for this information. For example, earlier analyses using the USDA Forest Service, National Woodland Owner Survey [
5] suggest that easements on family forest lands specifically may almost be twice as prevalent as those found using the Protected Areas Database.
There are also finer level ownership attributes, such as ownership and management objectives, financial resources, and size of holdings, which are not captured by the C&I, but this is understandable given that the C&I were developed for country level assessments. Relevant differences in ownership/management objectives include, for example, more of a focus on economic returns for Corporate forestland [
44] and more of a focus on amenity values for Family forestland [
1], but there is a mix or continuum of objectives within each category, and this diversity can be advantageous in that a broader mix of goods and services are more likely to be provided.
Another area influenced by the legal and institutional frameworks is the awareness levels of owners and the tools and resources available to them. There are several hypotheses for why Family forest ownerships may trail other ownership categories in BMP implementation, including lower awareness of BMPs and their importance [
47], many years or decades between harvests, which decreases owners’ exposure to these requirements and recommendations [
48], and a lack of technical resources [
48].
The Montréal Process frames legislation and policies that conserve and support forest function as a prerequisite to sustainable forest management. For example, secure land tenure is important for the welfare of people and forests. While most Private and Public ownerships have secure title, the history of land tenure for Indigenous people is wrought with land seizures and problematic treaties. Despite these challenges, Tribes have been noted for their exemplary forestry practices, and an increase in self-governance and self-determination has strengthened their ability to foster healthy forests [
49]. Although outside the scope of this paper, there are unique histories and policies related to the rights of Indigenous peoples in Alaska, Hawaii, and the protectorates and territories of the USA that are important to consider in the context of forest sustainability in those areas.
While tenure is secure for most Family forest ownerships, this is not the case for the forestland that is held as heirs’ property, where land has been passed on to future generations without clear title to the land and this creates challenges for owners’ abilities to manage and maintain ownership of their forestland [
50]. Heirs’ property is most common among Black landowners in the South, poor rural landowners in Appalachia, and Indigenous landowners [
51,
52].
Just over half of the Montréal Process C&I, i.e., Criteria 6 and 7, are socioeconomic in nature, but the C&I summarized in this article are largely biophysical in nature. This is due to data limitations. For example, logging and wood products represent only a portion of the economic benefits provided by the forests, which also provide many other ecosystem services, such as clean water, recreational opportunities, and nontimber forest products, but these are more difficult to directly measure, let alone separate by ownership category.
Assessing sustainability is fraught with challenges from how to define it, to how to measure it, to how to interpret the results. The Montréal Process C&I while imperfect, as are all systems designed to assess abstract concepts, have been developed by a broad coalition, have been widely applied, and have been periodically revised [
7,
53]. For objectives like those of this paper, a comparison across ownership groups, the C&I provide an objective means of assessing commonalities and differences through syntheses of data that are often complex and disparate. However, the Montréal Process C&I were not designed with this objective in mind, and some metrics can therefore be harder to interpret, such as when forest acreage is “lost” from one ownership group to another. The application of C&I necessitates examination of broad patterns, which are important in showing broad trends, and hopefully highlight areas where the finer details can be expounded on in future efforts and causal factors investigated.
The data and methods available for assessing forest sustainability are continually changing [
54]. The USDA Forest Service, Forest Inventory and Analysis program is the primary data source used here and in most assessments at the national, regional, and state level in the USA. As with national forest inventories for other countries, the consistency of data collection over time is critical. Harmonization of data collection will facilitate comparisons and is what made the cross-ownership comparison of this paper possible. At the same time, there are many ecosystem services, such as recreational access, and socio-psychological factors, such as ownership and management objectives, that are not regularly assessed nor reported across all ownership categories.
The underlying trends in forest sustainability are influenced by many factors, some of which are in the immediate control of the owners and others that are not. Historical factors, including colonization, homesteading, land forfeiture, and land acquisition acts, such as the Weeks Act which was responsible for the creation of most of the National Forests in the eastern USA, have influenced the broad ownership patterns across the USA. This in turn is connected to forest productivity, forest types, and other factors.
Although the focus of this paper is on the USA, the findings are potentially applicable to other countries with similar ownership types. What is limiting international comparisons is a lack of analogous analyses in other countries where the Montréal Process is being implemented, but the analytical framework developed here could be applied. These analyses would allow for those countries to compare trends across ownership categories and, as with the results here, be used to inform policies and other approaches tailored to the opportunities and challenges confronted by different ownership groups.