Definition and Classification

While the definition of “non-timber forest product” can vary, most literature on NTFPs in the United States focuses on specific plants or fungi that grow in forests. NTFPs do not include “timber,” that is, wood utilized at industrial scales or sold in commodity markets, but they do include firewood and wood for artisanal purposes (Belcher Reference Belcher2003; Chamberlain, Emery, and Patel-Weynand Reference Chamberlain, Emery and Patel-Weynand2018). Most literature on NTFPs in the United States does not include animal wildlife, which typically receives separate treatment in forest management and scientific literature.

Five market “segments” define the NTFP sector (Chamberlain et al. Reference Chamberlain, Davis, Duguid, Ellum, Farrell, Friday, Fuller, Kruger, Marcano-Vega, McCoy, Page, Prisley, Young, Zasada, Chamberlain, Emery and Patel-Weynand2018): edible products, medicinal and dietary supplements, decorative products, landscaping inputs, and supplies to arts and crafts. Examples of each segment illustrate the diversity of products and markets. Fruits, berries, nuts, sap, fern fronds, tubers and bulbs are edible products. Each year, sales of maple (Acer spp.) syrup contribute more than $100 million to the U.S. economy (Farrell and Chabot Reference Farrell and Chabot2012). Thousands of kg of ramps (Allium tricoccum), a native forest onion, are harvested each year for the expanding foraged food market (Baumflek Reference Baumflek2016). The medicinal and dietary supplement segment of the NTFP industry depends on large quantities of plant materials harvested from forests for their therapeutic values. Extract of Pacific yew (Taxus brevifolia) bark was found in the early 1990s to be effective for treating ovarian cancer (Chamberlain, Bush, and Hammett Reference Chamberlain, Bush and Hammett1998). Forest plants and their parts are sold to the decorative segment as greenery in floral arrangements or as the main component of fresh and dried ornaments. In 1995, the United States exported more than $14 million in forest-harvested mosses and lichens, most of which originated from Appalachia and the Pacific Northwest (Goldberg Reference Goldberg1996). Live forest plants are also collected for the landscaping industry. At one time, seedlings of Fraser fir (Abies fraseri), endemic to the Southern Appalachians, were “pulled” from the forest floor for transplanting to Christmas tree farms. Collection of forest understory wildflowers for horticultural sales is a cottage industry in parts of the eastern United States (Cullina Reference Cullina2000). The number of NTFPs used for arts and crafts is limited only by the imagination of the artisan (Chamberlain et al. Reference Chamberlain, Davis, Duguid, Ellum, Farrell, Friday, Fuller, Kruger, Marcano-Vega, McCoy, Page, Prisley, Young, Zasada, Chamberlain, Emery and Patel-Weynand2018, p. 17). In general, though, NTFP extraction and sale for markets has been little studied relative to other forest goods and services, and the economic importance of these natural resources have, therefore, been underappreciated.

Methods of Production and Sale

Typically, NTFPs are sourced through two types of production methods. In traditional “wild harvesting,” people forage the products from forests, with few inputs except for their time, transport, and minor incidentals such as a digging tool. “Forest farming,” by contrast, is the cultivation of forests to produce NTFPs (Chamberlain et al. Reference Chamberlain, Davis, Duguid, Ellum, Farrell, Friday, Fuller, Kruger, Marcano-Vega, McCoy, Page, Prisley, Young, Zasada, Chamberlain, Emery and Patel-Weynand2018). Several methods of forest farming are generally accepted (Chamberlain et al. Reference Chamberlain, Mitchell, Brigham, Hobby, Zabek, Davis and Garrett2009). The “wild-simulated” forest farming system includes only limited site preparation and planting, whereas a “woods cultivated” forest farming system is more intensive, potentially incorporating tillage and soil amendments, and “woods grown” forest farming methods rely on more inputs and maintenance. The method of production clearly has direct implications for production costs, and thus the valuation of the provisioning ES.

The level of market formality and the method of sale also have implications for ES valuation. In a first case, the landowner and harvester may be one and the same. Second, the harvester may access sites without compensating the landowner (or indeed, in some cases, without the landowner's knowledge or permission). In this case, the first point of sale may be between harvester and dealer. Third, the harvester may compensate the landowner, but outside competitive markets. This may be particularly true for harvest permits on federal lands, such as the National Forests. Fourth, commercial harvesters may directly pay a landowner for access in a competitive marketplace, as seems to be true in certain instances for saw palmetto (Serenoa repens) in Florida, and pine (Pinus spp.) straw in Georgia. Fifth, the harvesters may be employees or contractors of a larger company and may be paid either per hour or per volume of product collected. Lastly, in many cases which are important for future work but not considered in this manuscript, no market transaction takes place at all. Rather, the product is directly consumed by the harvester or given to friends and family (Frey et al. Reference Frey, Blatner, Jacobson, Downes, Sills, Mercer, Alexander, Chamberlain, Gold, Godsey, Emery, Coffin, Barlow, Lohr, Chamberlain, Emery and Patel-Weynand2018).

Conceptual Model

Many users directly harvest NTFPs for personal subsistence, recreational, cultural, or spiritual value rather than for commercial sale. In these cases, parallels to recreation demand are clear, and traditional revealed or stated preference non-market valuation methods can be applied. The travel cost method, in particular, has been used for NTFP ES valuation (e.g., Markstrom and Donnelly Reference Markstrom and Donnelly1988; Starbuck et al. Reference Starbuck, Alexander, Berrens and Bohara2004). While important tools, such methods are not the focus of this study.

Derived demand

The demand for a market-based NTFP provisioning service is a “derived demand,” in the sense that it is an input to the production of a good for sale (Flaaten Reference Flaaten2011, p. 31; McConnell and Bockstael Reference McConnell, Bockstael, Mäler and Vincent2005). We focus here on how best to estimate the provisioning ESV at the margin, which can be described as the resource quasi-rent (ρ) from the natural resource to the human owner of the service (Freeman Reference Freeman2003; Kronenberg and Hubacek Reference Kronenberg and Hubacek2013; McConnell and Bockstael Reference McConnell, Bockstael, Mäler and Vincent2005):

(1)$$\rho = R-VC = [ {\,p\cdot q\lpar {l\comma \; k\comma \; e} \rpar } ] - [ {w\cdot l + v\cdot k} ] $$

where

ρ is the resource quasi-rent, or ESV, equal to revenue (R) minus variable costs of production (VC). R is equal to price (p) times quantity (q).

q(l, k, e) is the production function, which maps quantity harvested as a function of the inputs labor (l), capital (k), and the ecosystem service (e).

w and v are input prices.

To estimate the value of non-marginal changes of the ES, the production function q(·) must be known at all production levels and for all producers. Further, if the change in the ES ultimately induces a change in the price (p) of the consumer good, then, to fully estimate the changes in welfare, understanding the demand function for end-use consumer goods will be necessary (Freeman Reference Freeman2003; McConnell and Bockstael Reference McConnell, Bockstael, Mäler and Vincent2005). In practical terms and at the marginal level, the resource quasi-rent or provisioning ESV may be thought of as that portion of product value that is owed to ecosystem function, or the “residual value” after factoring out the value added through human labor and capital (Smith, Crone, and Alexander Reference Smith, Crone and Alexander2010).

Access to a resource affects who can harvest and how much they can take. In the extreme case, open access resources allow any harvester to take as much as desired, as appears to be the case for certain marine fisheries (Gordon Reference Gordon1954; Flaaten Reference Flaaten2011), and, arguably, in a de facto sense for some NTFPs, when access is difficult to control and enforce (Bulte and Engel Reference Bulte, Engel, López and Toman2006; Frey, Chamberlain, and Prestemon Reference Frey, Chamberlain and Prestemon2018). In such situations, economic theory holds that economic overproductionFootnote ¹ will drive resource quasi-rent to zero (Flaaten Reference Flaaten2011). If the landowner can control access, a harvest level lower than the open access equilibrium can be enforced, generating resource quasi-rents.

Net present value

We define provisioning ESVs (resource quasi-rents/residual values) in terms of dollars per unit of area by forest type—perhaps a bit of sleight of hand, as the area of forest is not the service itself, yet is a reasonable simplification for the purposes of use in forest management choices. In the simplest case, if a price is paid directly from the harvester to the landowner per unit area of forest accessed, then the contract price is equal to R (equation 1). If the landowner has no establishment or management costs (the case for a wild-harvested product), then the price per unit area of the contract is equal to the provisioning ES (ρ _t) for a given year t:

(2)$$\rho _t = R_t$$

However, the situation is usually not so simple with NTFPs. In many cases, the first point of sale is not the harvester who pays a landowner per unit area for the right to access and harvest a product in the forest, but rather the harvester receiving payment from a dealer per unit of product. In some cases, permit fees are not set in a competitive marketFootnote ² and may not reflect the true ESV. With such a non-market-based permit, or when no fee is paid to the landowner, the analyst should use the first point of sale (P ₁) in the competitive marketplace for valuation, taking into account that the harvester may have added value to the product through harvesting, transacting (transporting, marketing, selling), and possibly processing (e.g., cleaning, drying) the product. These human costs must be factored out to accurately estimate residual price (P ₀), which is the shadow price of the product in the woods:

(3)$$P_0 = P_1-\lpar {HC + TC + PC} \rpar $$

(4)$$\rho _t = R_t = \lpar {P_{1\comma t}-\lpar {HC_t + TC_t + PC_t} \rpar } \rpar \cdot Q_t$$

where

P ₁ is the market price per unit of product at the first point of sale,

P ₀ is the residual or shadow price per unit of product in the woods,

HC and TC are the harvesting and transaction costs per unit of product, including time traveling, collecting, marketing, and selling the product, plus fuel for transportation, depreciation of equipment, vehicles, and other factors,

PC is the processing cost per unit of product of value-adding activities taking place before the first point of sale, including cleaning, drying, and

Q _t is the quantity of NTFP produced in year t on a marginal unit of forestland of a given forest type and ownership class.

Forest farming and managing natural populations may incur costs associated with establishment and management, which occur before harvest. These are variable costs but are usually associated with costs per unit of land area rather than costs per unit of product:

(5)$$\rho = R_0-VC = [ {\lpar {P_1-\lpar {HC + TC + PC} \rpar } \rpar \cdot Q} ] - [ {EC + MC} ] $$

where EC and MC are establishment and management costs. However, because establishment and management costs occur many months or even years before harvest, analysis should consider the opportunity cost of time by discounting future values relative to present ones. The quasi-rent, or residual value, over a finite period T, considering the opportunity cost of time, is given by net present value (NPV), which can be converted to the soil expectation value (SEV) for an infinite time horizon or the equivalent annual income (EAI) (Alexander et al. Reference Alexander, Pilz, Weber, Brown and Rockwell2002; Cubbage et al. Reference Cubbage, Davis, Frey, Behr, Sills, Pancel and Kohl2015):

(6)$$\eqalign{\rho _T & = NPV = \mathop \sum \limits_{t = 1}^T \displaystyle{{R_t-C_t} \over {{\lpar {1 + \delta} \rpar }^t}} \cr & = \mathop \sum \limits_{t = 1}^T \displaystyle{{[ {\lpar {P_{1\comma t}-\lpar {HC_t + TC_t + PC_t} \rpar } \rpar \cdot Q_t} ] - [ {EC_t + MC_t} ]} \over {{\lpar {1 + \delta} \rpar }^t}}}$$

(7)$$\rho _\infty = SEV = NPV + \displaystyle{{NPV} \over {{\lpar {1 + \delta} \rpar }^T-1}}$$

(8)$$\rho _t = EAI = SEV\cdot \delta $$

where

NPV is the per unit area net present value over the finite “rotation” period T, on forestland of a given forest type and ownership class, and

δ is the discount rate.

Geospatial variation

The harvestable quantity of a product is not constant over space but varies according to population characteristics and site conditions. These differences are critical in marginal valuation since site production varies. Some sites (i) will support more harvest (Q) than others as:

(9)$$Q_{t\comma i} = q\lpar {H_i\comma \; Q_{t-p\comma i}} \rpar $$

where H is a vector of habitat characteristics affecting growth and yield of the plant, and Q_t-p is harvest in a past period, t-p.

Furthermore, harvest and transaction costs (HC and TC) vary based on numerous factors as:

(10)$$TC_i = tc\lpar {D_i} \rpar $$

(11)$$HC_i = hc\lpar {G_i\comma \; S\comma \; \; Q_i} \rpar $$

where

D _i is a vector of variables representing distance from market or population center from access point and distance and terrain from access point to harvest site.

G is a vector of variables representing the difficulty of the topography and terrain at the harvest site,

S is a vector of variables representing species characteristics affecting the difficulty of harvest, and

These functions must be defined and integrated into the relevant equation (4–6), for a given species over multiple sites.

Challenges in Valuation

Several related challenges lead to difficulties obtaining data and applying research at broader spatial and temporal scales.

Past Valuation Studies in the United States

Few studies fit the paradigm of NTFP provisioning ES valuation. Among those using the NPV framework (equations 6–8) for NTFPs are those for wild mushrooms in the Pacific Northwest, which assume joint production of mushrooms and timber (Alexander et al. Reference Alexander, Pilz, Weber, Brown and Rockwell2002; Pilz et al. Reference Pilz, Brodie, Alexander and Molina1998; Pilz et al. Reference Pilz, Smith, Amaranthus, Alexander, Molina and Luoma1999). A particular challenge in this system has been estimating harvest, transaction, and processing costs (HC + TC + PC in Equations 3–6). These include time scouting and traveling to harvest sites, harvesting, cleaning, and selling the product (Love, Jones, and Liegel Reference Love, Jones and Liegel1998). To resolve this challenge, the authors assumed costs either to be a fixed percentage of the sale price or a range of estimates (Alexander et al. Reference Alexander, Pilz, Weber, Brown and Rockwell2002; Pilz et al. Reference Pilz, Brodie, Alexander and Molina1998; Pilz et al. Reference Pilz, Smith, Amaranthus, Alexander, Molina and Luoma1999). Depending on species, site, and assumptions about harvest costs, management, and productivity, estimated SEVs for mushrooms have ranged widely, from $17/ha to $2,478/ha (Alexander et al. Reference Alexander, Pilz, Weber, Brown and Rockwell2002) at a 4% discount rate. In some cases, timber value was 10 to 100 times greater than mushroom value, while in other analyses, where timber values were low and mushroom values high, mushroom value equaled or exceeded timber value (Alexander et al. Reference Alexander, Pilz, Weber, Brown and Rockwell2002; Pilz et al. Reference Pilz, Smith, Amaranthus, Alexander, Molina and Luoma1999).

Forest farming incurs establishment and management costs (EC + MC in equations 5–6). Burkhart and Jacobson (Reference Burkhart and Jacobson2009) estimated NPVs for forest farming of medicinal plants in the eastern United States and, with the exception of American ginseng, found them to be unprofitable. Determining an appropriate hourly cost for labor (wage rate) and a discount rate, as well as projecting product prices (which can be volatile) into the future, proved particularly challenging. Wage rate was estimated using an opportunity cost of $13/hr, based on average wages for comparable types of work. A range of possible values was used for discount rate and product price.

Define the geographic area

NTFPs are harvested from most forest types across every region of the United States. Defining geographic boundaries and setting priorities as to products for estimating ESV are essential first steps (Chamberlain et al. Reference Chamberlain, Frey, Ingram, Jacobson, Downes, Sills, Moore, Cubbage, McCarter, Holmes and Mercer2017). Some forest types are more productive and thus warrant higher priority for ES valuation. For example, Appalachian hardwood forests have a high diversity of plants harvested for food and medicine, supplying large quantities of raw materials to industry. These forests span most eastern states, and estimating ESVs for the entire region is probably unrealistic. A better approach may be to focus on a particular state, forest type, or forested area. With more geographic specificity, it is more feasible to identify and evaluate sites known as sources of NTFPs (e.g., a well-known ramp or blueberry patch).

Recognize forest type, ownership, and accessibility

Within geographic areas of interest, understanding forest types, the habitat requirements of the NTFP, and factors related to harvester access is critical. Valuation efforts should also consider the resource base as it interacts with changing ownership and management (Chamberlain et al. Reference Chamberlain, Frey, Ingram, Jacobson, Downes, Sills, Moore, Cubbage, McCarter, Holmes and Mercer2017), which may affect NTFP availability and ESVs. For example, if the availability of NTFPs on public forests decreases due to increased regulation, demand for NTFPs from private lands might increase, thus increasing ESVs on those lands. The relative importance of land ownership type varies across the United States, as western states have more public lands and eastern states more private lands. Conversely, as private forest habitats are impacted by land use change (e.g., mountain top removal, timber cutting, residential development), the ESV from other ownerships will be affected.

Identify NTFPs for ES valuation

Deciding which NTFP ESV to estimate is a crucial step. Institutions have limited resources for estimating the ESV and must therefore be selective. Product demand, ecological vulnerability, the status of the geographic area in the broader market, and cultural importance are key criteria (Chamberlain et al. Reference Chamberlain, Frey, Ingram, Jacobson, Downes, Sills, Moore, Cubbage, McCarter, Holmes and Mercer2017; Emery and Ginger Reference Emery and Ginger2014).

Priority NTFPs may be those well-known locally, nationally, and internationally, for which demand has increased, or those vulnerable to overharvesting or other threats. For example, it may be important in some states to estimate the ESV of wild American ginseng, given the current high demand in Asia and potential for overharvest (Ticktin Reference Ticktin2004). The plant part (e.g., roots vs. berries) harvested may increase the potential for population declines from harvesting (Ticktin Reference Ticktin2004), increasing priority for valuation.

Actual or potential harvest quantities in an institution's geographic area of interest/influence, relative to other regions, may indicate to decision-makers that valuation is worthwhile. This perception of importance may reflect actual market demand for the product(s), as well as standing stocks of species in the forest. If the standing stock is not known, estimation of the habitat for important NTFPs within the geographic area of interest may indicate relative potential for producing priority NTFPs. The uniqueness of a product and the market stature of a producer may factor into the decision as well. For example, the state of Florida is the major producer of saw palmetto, so it makes sense that this product would be a priority for that state, as opposed to Georgia, which has saw palmetto growing but is not a major producer. Conversely, Georgia is a major producer of pine straw, which makes this more of a priority for that state.

In setting priorities for product valuation, institutions may consider the economic and cultural importance of certain products to specific communities within a geographic area of interest. The perceived importance of a product to a region's economy, or to a specific segment of the region's population could be a justification for estimating the ESV of that product. For example, annual ramp festivals are important to the economy of western North Carolina. As such, estimating the provisioning ESV of this edible forest product may be warranted.

Understand method of production and sale; harvest site access; data availability

Before beginning the valuation process, recognize that the valuation method will depend on the method of production (typically wild-harvested or forest farmed) and sale, market chains, land ownership and access, size of the geographic area of interest, as well as data availability. For truly wild-harvested products, no establishment or management costs exist—people forage for products in forests that are not managed for that product. For woods-grown, woods-cultivated, or wild-simulated products, relevant establishment and management costs must be considered. For products from managed natural populations, establishment costs may be negligible, but there could be management costs.

Production Costs

We identified five main types of costs (establishment, management, harvest, processing, transaction) in equations (5–6). These costs may vary between harvesters and products. Establishment and management costs for forest-farmed products are typically associated with the landowner, given as costs per unit area. Woods-cultivated and woods-grown costs are higher than those of wild-simulated. Establishment costs are zero for managed natural populations, and both establishment and management costs fall to zero for wild-harvested products. Harvesting, processing, and transaction costs are associated with the harvester, and are given as costs per unit of product sold.

These categories provide a broad guide. A full accounting of relevant costs is needed to allocate expenses appropriately. The first step in estimating landowner and harvester costs is to identify and describe those costs. Understanding the actual processes, activities, and timing of the costs is critical (Cubbage et al. Reference Cubbage, Davis, Frey, Behr, Sills, Pancel and Kohl2015). Magnitude of inputs for each activity can then be estimated, including hours of labor, quantities of supplies, services contracted, and equipment used. Finally, a price for each input must be determined to estimate the full cost.

Many commercial harvesters and landowners are self-employed and do not pay themselves per hour worked, and may also employ unpaid family labor. These producers may not consider their time to be a “cost” in the financial sense. However, including this economic cost is essential when estimating a provisioning ESV. Human labor represents a cost distinct from the ecosystem input.

Permit or contract fees for access to the resource are not among the costs in equation (5), because permits represent a simple transfer from harvester to the landowner rather than an economic cost of production. In some instances, this transfer itself could be an estimate of the ESV. For example, if the permit fee or contract price to access and extract the product are set in a competitive marketplace, such as when private landowners sell the right to access and harvest saw palmetto or pine straw on their land, the price may represent the full quasi-rent (equation 2) or an estimate of the costs of establishment and management. On public lands, federal agencies may base permit fees on an estimate of the ESV per unit of product harvested (Smith, Crone, and Alexander Reference Smith, Crone and Alexander2010),Footnote ³ or the permit fee may offer a lower bound for total ESV.

Harvest Quantity

To estimate the provisioning ESV of NTFPs requires that the volume harvested per unit area of habitat be determined. In general, this is done by (1) estimating standing stock volume of product in the forest, or (2) estimating rate of harvest (i.e., flow) from a defined forest area. Standing stock measures the potential for future harvests. However, estimating the ESV of standing stock is hypothetical by nature because harvesting a large volume of the standing inventory at once could impact market prices and the sustainability of the resource. Moreover, determining the standing stock of NTFPs is challenging, particularly for products harvested for their roots, as there are few methods (Chamberlain et al. Reference Chamberlain, Ness, Small, Bonner and Hiebert2013).

The most straightforward approach, and one generally favored by researchers, is to estimate the rate of harvest from a specific site in a forest. The flow approach quantifies the raw material removed from an area over a period and accounts for the volume that reaches market. The best data is collected at the location of harvest, although this requires being on-site during harvest or having a relationship with trusted harvesters who will report accurately. Permitted harvest amounts can be a starting point, recognizing that permit holders may harvest less than (or on occasion, more than) their permit allows. As noted previously, one challenge with the flow approach is that rates of harvest may be unsustainable.

The approach to estimating quantity produced will depend on the forest area defined. Ideally, a specific harvest site of fixed area has been identified where the quantity removed can be closely tracked. If this is not the case, then precise characterization of the forest type and habitat for the species of interest must be determined. Variables such as soils, elevation, slope, and aspect may be used to quantify habitat quality (Schmidt et al. Reference Schmidt, Cruse-Sanders, Chamberlain, Ferreira and Young2019). The analysis should also include factors related to human access, such as trails, roads, and proximity to harvester residences. The more precisely the site can be described, the more accurate the estimation of the area of potential forest habitat for a particular NTFP or suite of products. On-the-ground verification can further improve the accuracy of estimations.

Price

As mentioned, price of NTFPs at the location of harvest will lead to the most accurate ESV estimate. It is possible that sales occur at different places along the market chain, with some enterprises buying only from harvesters and others buying from harvesters and other dealers (Kruger Reference Kruger2018). This can bias data, as it opens opportunities for double-counting harvest volumes. The farther along the market chain that price is collected, the more opportunities for inaccuracies are present due to additional steps of value adding. Realistically, price data are collected at the first point of sale, when the harvester sells the product to a dealer. This first point of sale, though, includes value adding through harvesting, cleaning, and transport.

Existing Sources

Estimating production costs requires understanding of production methods, activities, and inputs, which to our knowledge are not tracked by any data service. Past research that includes estimates of production costs has elicited this information from expert knowledge, interviews with harvesters, or monitoring of forest farmer activities (e.g., Burkhart and Jacobson Reference Burkhart and Jacobson2009; Davis and Persons Reference Davis and Persons2014; Huyler Reference Huyler2000), and cooperative extension services sometimes produce “enterprise budgets” that provide a valuable reference point (e.g., Frey Reference Frey2014). These offer cost estimates but are generally neither updated nor replicated at multiple geographic locations.

Once activities and inputs are known, a dollar value must be placed on each. The U.S. Bureau of Labor Statistics provides data on average wages by occupation, which can be used to find an appropriate opportunity cost of time for a typical harvester by geographic area (Bureau of Labor Statistics 2019). The Internal Revenue Service provides standard rates that can be used to estimate travel costs from the forest to the first point of sale.

Few sources provide data on the quantity of NTFPs harvested. The main sources of harvest volumes from federal forest lands are the USDA Forest Service's “cut-and-sold” reports, and similar reports from the Bureau of Land Management (Chamberlain, Teets, and Kruger Reference Chamberlain, Teets and Kruger2018). These reports record permitted harvest volumes but not necessarily actual harvest volumes. Data reported at the forest level may be tracked to the district level but not disaggregated by habitat type. Thus, estimates of habitat area are needed to complement harvest data from reports. Through the USDA Forest Service's Forest Inventory and Analysis program, tree data is available from which the inventory of some NTFPs can be estimated. For example, trees harvested for bark are inventoried, and this data can be used to estimate the volume of bark harvested across the range of the species (Kauffman, Prisley, and Chamberlain Reference Kauffman, Prisley and Chamberlain2015). The USDA Forest Service's National Woodland Owners Survey provides data on the proportion of private landowners who report the collection of NTFPs on their property but does not give species, commercial harvest volume, or price (Butler et al. Reference Butler, Hewes, Dickinson, Andrejczyk, Butler and Markowski-Lindsay2016a, Reference Butler, Hewes, Dickinson, Andrejczyk, Butler and Markowski-Lindsayb).

Other federal and state agencies collect data on specific NTFPs. The National Agricultural Statistics Service (NASS) collects information on production of maple syrup, blueberries, and other well-known commodities, and these data may be used to estimate harvest volumes for states or regions (e.g., NASS 2018). The U.S. Fish and Wildlife Service, the federal authority for implementation of the Convention on International Trade of Endangered Species of Wild Flora and Fauna (CITES) that regulates American ginseng exports, is the repository of data on harvests reported at the state and county level. Some states allow harvesting on their forest lands and may have harvest permit information for some of the more common NTFPs (Frey and Chamberlain Reference Frey and Chamberlain2016). Georgia, for example, provides data on pine straw production through the farm gate value reports (e.g., Wolfe and Stubbs Reference Wolfe and Stubbs2013, Reference Wolfe and Stubbs2015).

Trade associations, such as the American Herbal Products Association (AHPA), collect information on harvest volume and price for specific products (e.g., Dentali and Zimmermann Reference Dentali and Zimmermann2012). University, government, and independent research programs from time to time have tracked input cost, harvest volume, and price for NTFPs (e.g., Alexander et al. Reference Alexander, Pilz, Weber, Brown and Rockwell2002; Davis and Persons Reference Davis and Persons2014; Greenfield and Davis Reference Greenfield and Davis2003; Huyler Reference Huyler2000; Kruger Reference Kruger2018; Pilz et al. Reference Pilz, Brodie, Alexander and Molina1998; Schlosser and Blatner Reference Schlosser and Blatner1995), typically relying on interviews with dealers and harvesters over limited periods of time.

Surveys and Interviews

Because existing sources of data related to production costs, harvest quantity, and price of NTFPs are generally lacking, it is often necessary to interview market players. The closer the data source is to the point of harvest, the greater the precision of the data. As those closest to the ES, harvesters are the preferred sources of information. The most statistically defensible approach is to randomly sample as large a subpopulation from the full population of harvesters as possible. Permits on public lands may be a source of information about harvesters, who can then be sampled for a survey. Starbuck et al. (Reference Starbuck, Alexander, Berrens and Bohara2004), for example, compiled the names and addresses of all holders of recreational use special forest product harvesting permits from two National Forest districts in 1996 and conducted a mail survey of a sample of this population to estimate consumer surplus based on the travel cost method of non-market valuation. A similar approach could be used to contact harvesters for information related to market-oriented uses and to elicit information on production cost, harvest quantity, and price.

Often there is no list of harvesters, so random sampling is impossible. In this case, using expert knowledge to identify key groups of individuals and selecting a robust number from each stratum may be the best option. Failing a stratified sampling approach, a chain-referral approach may be used until saturation of the population or a substantial sample has been reached (Chamberlain et al. Reference Chamberlain, Frey, Ingram, Jacobson, Downes, Sills, Moore, Cubbage, McCarter, Holmes and Mercer2017).

In certain instances, reaching harvesters is not possible, and the first point along the market chain when the products enter the formal market may be the most feasible point to sample buyers. In the case of medicinal plants in the eastern United States, the list of registered ginseng buyers has proved an excellent source for identifying buyers of other medicinal plants (Kruger Reference Kruger2018), but it covers only 19 states.

A survey of wholesale firms in a specific segment of the industry may provide insights into ways to access harvesters. A snowball sample of market participants will help to identify the majority of firms. A stratified sample of known market segments (e.g., wholesale distributors of food or floral products) may be possible by examining databases available through government agencies.

Saw palmetto

Saw palmetto, typically found in the understory of pine forests, is endemic to the Coastal Plain from South Carolina to Louisiana. The berries of the saw palmetto are used for medicinal purposes with an estimated average of approximately 8 million kg/yr (18 million lbs/yr) of fresh berries harvested—almost exclusively from Florida (Dentali and Zimmermann Reference Dentali and Zimmermann2012). ‘Independent’ harvesters were paid $0.22–0.24/kg ($0.10–0.11/lb) of fresh berries (Carrington et al. Reference Carrington, Mullahey, Krewer, Boland and Affolter2000). Mitchell (Reference Mitchell2014) recorded prices paid to harvesters ranging from $0.22–2.20/kg ($0.10–$1.00/lb), noting that on average harvesters were paid $1.10/kg ($0.50/lb) of fresh berries. Landowners who hired crews to harvest received $0.44–0.46/kg ($0.20–0.21/lb) of fresh berries, and paid harvesters about $0.15/kg ($0.07/lb) (Carrington et al. Reference Carrington, Mullahey, Krewer, Boland and Affolter2000). The price of harvester labor is clearly dynamic, and, although challenging, accurately estimating labor costs requires the active monitoring of buying stations during the harvest season and direct communication with harvesters to determine how much they get paid.

However, the saw palmetto market is structured such that, in many cases, a harvest contractor will pay a landowner directly, in a competitive marketplace, per unit area per year for the right to harvest saw palmetto berries from their forest. In this instance, the value of the contract between the harvester and landowner already has factored out the harvester's costs, and if the landowner has not invested any resources in establishing or managing the saw palmetto in the forest, the contract value can be considered equivalent to the quasi-rent or ESV at the margin. Carrington et al. (Reference Carrington, Mullahey, Krewer, Boland and Affolter2000) reported harvest contract values ranging from $38–77/ha/yr ($15–31/ac/yr), which would be the EAI and translates to an SEV of $760–$1,540 per hectare ($308–625/ac) at a 5% discount rate (equations 2, 6–8).

American ginseng

American ginseng is an understory plant found in eastern hardwood forests. The roots are highly valued, particularly in East Asian cultures, for a variety of medicinal purposes. Because of a long history of widespread (destructive) harvest of roots, American ginseng is a species of conservation concern and listed under CITES, and therefore harvest for export is tracked by the U.S. Fish and Wildlife Service (2015). Annually, an average of approximately 27,000 kg dry (60,000 lbs dry) of wild American ginseng are harvested from eastern forests for export (Chamberlain, Prisley, and McGuffin Reference Chamberlain, Prisley and McGuffin2013; Frey, Chamberlain, and Prestemon Reference Frey, Chamberlain and Prestemon2018).Footnote ⁴

Approximately 0.5% of the American ginseng harvest comes from National Forests in Georgia, Kentucky, North Carolina, and Tennessee.Footnote ⁵ Let us consider the Pisgah and Nantahala National Forests in North Carolina (NFSNC) as a case study.Footnote ⁶ The NFSNC include 422,670 ha (1,044,000 acres) in the southern Appalachian Mountains. The USDA Forest Service cut and sold reports indicate that from 2009 to 2012, an average of 260 kg/yr (573 lbs/yr) of green American ginseng root were permitted for harvest from NFSNC (Table 1) (Chamberlain et al. Reference Chamberlain, Frey, Ingram, Jacobson, Downes, Sills, Moore, Cubbage, McCarter, Holmes and Mercer2017).

Table 1. First approximation of average value estimates of American ginseng on the National Forests in North Carolina, 2009–2012.

^a Source: Adapted from Davis and Persons (Reference Davis and Persons2014).

^b American ginseng on the NFSNC is assumed to be wild-harvested, so there is no establishment or management cost. We include the column as a reminder to the reader.

Davis and Persons (Reference Davis and Persons2014) interviewed ginseng buyers (the first point of sale after harvest) throughout the eastern United States each year between 1982 and 2012 during harvest season and reported a range of prices. From 2009 to 2012, the average midpoint price was $1,411/kg dry ($640/lb dry, Table 1), translating to approximately $470/kg green ($213/lb green).

As a first approximation, we estimated approximate average ESVs for ginseng in the NFSNC. We assumed that National Forest permits were equal to harvests, that they were entirely from unmanaged populations where no resources were invested in either establishment or management, that the equipment and supplies used by harvesters were of negligible cost, and that the labor of harvesting, transacting, and processing roots was the only non-negligible production input and equaled 8 hrs/kg green at an opportunity cost of $10/hr.

We used equation (5) to estimate an annual average value for American ginseng on the NFSNC (Table 1). Because the price and quantity could be correlated either positively or negatively, we did not simply multiply the annual average price minus cost by the annual average quantity harvested. Rather, we calculated the ESV as an average EAI of $94,462/yr, or an SEV of $1,889,230 at a 5% discount rate, translating to $4.47/ha ($1.81/ac) or $0.22/ha/yr ($0.09/ac/yr).

To be actionable for forest management decision-making purposes, more spatially explicit information on ginseng occurrence and harvest size needs to be included in the valuation model. Thus, to translate estimates of average value per ha into estimated marginal values per ha, we incorporated proxies for spatial variability in harvestable quantity and for harvest costs. We assumed that the probability of ginseng presence was determined by habitat factors, including parent material, landform, climate, and soil taxonomy, as outlined in Schmidt et al. (Reference Schmidt, Cruse-Sanders, Chamberlain, Ferreira and Young2019). We used a raster (9 ha [22 acre] pixel resolution) of probabilities of ginseng occurrence generated by the species distribution model (SDM) developed by Schmidt et al. (Reference Schmidt, Cruse-Sanders, Chamberlain, Ferreira and Young2019). After clipping the SDM raster to the boundary of the NFSNC, we binned probabilities of finding ginseng on a given pixel as 0.1 (23,695 pixels), 0.5 (18,335 pixels), and 0.9 (3,089 pixels).

Although actual presence and total quantity of ginseng harvestable at the pixel are unverified, the purpose of this exercise was to better understand the distribution of potential ESV within the NFSNC. First, using a random number generator and the probabilities just given, we randomly assigned ginseng presence or absence to each pixel. This leads to an average of 14,317 pixels with ginseng presence out of 45,149 total pixels, or 32%. Second, for each of the resulting pixels with ginseng presence, we then generated a random harvestable quantity from a beta distribution with mean 0.10 green kg and maximum of 0.60 kg per 9-ha pixel.Footnote ^7, Footnote ⁸ Third, we randomly assigned detection by harvesters each year at each presence pixel with probability of 0.5Footnote ⁹ and assumed that at each pixel where harvest occurred, 100% of the harvestable quantity was removed.

We modeled transaction and harvest costs as functions of distance to road, slope, and amount of ginseng present. Using the SDM raster as a template, we converted a shapefile of roadsFootnote ¹⁰ to a raster. Then, using the “distance” command in the “raster” package (Hijmans Reference Hijmans2019) in R (R Development Core Team 2005), we converted the road raster into a raster of distance (m) from road. To create a raster of slope (degrees) at the same resolution as the SDM data for the National Forest, we downloaded high-resolution elevation data (0.0225 ha) via the “elavatr” package (Hollister and Shah Reference Hollister and Shah2017) in R, and derived slope using the “slopeAspect” command. We estimated transaction and harvest costs for each pixel as

(12)$$TC_i = opp\cdot \left({\left({\displaystyle{{dist_i} \over {sp}}} \right)\cdot \lpar {1 + s_i} \rpar } \right)+ fc$$

(13)$$HC_i = \; opp\cdot \lpar {hr\cdot Q_i\cdot \lpar {1 + s_i} \rpar } \rpar $$

where

This simulation yielded an estimate of 425 kg/yr of green ginseng harvested with a total ESV of $88,978/yr, mean ESV across presence pixels of approximately AEI of $1.90 /ha/yr ($0.77/ac/yr) or SEV of $38/ha ($15/ac), and maximum value of approximately AEI of $11.50 /ha/yr ($4.65/ac/yr) or SEV of $230/ha ($93/ac).

These estimates are for the purpose of demonstration only and are highly uncertain. Although we do not estimate confidence intervals or quantify the level of uncertainty in the ESV estimates, additional research could surely improve these estimates. There are several sources of uncertainty in the underlying assumptions: (1) the estimated likelihood of ginseng presence based on ecological modeling, (2) the random assignment of harvestable quantity levels where ginseng was predicted to be present, (3) the likelihood of detection, (4) harvest costs, and (5) estimates of the first point of sale price, which were based on the best existing source (Davis and Persons Reference Davis and Persons2014), and provided information on the range but not the distribution of existing prices. Research to improve these estimates would likely require detailed interviews with many harvesters and buyers to generate more precise estimates of harvest quantities, costs, and prices. Further, our spatial methods only indicate likelihoods, not certainties, so improving estimates for individual sites’ field inventories would be needed.