In the O horizon, 0.5–4 mm charcoal C is 4 times that of 4–20 mm charcoal C (Table 1). In the mineral soil, <4 mm charcoal C is more than 30 times 4–20 mm charcoal C. Charcoal C concentration decreases from the surface (0–3 cm) to subsurface (3–15 cm) mineral soil (Table 1). Charcoal C is ~6% of total C in the O horizon of burned units, but <1% in unburned units (Table 1). In contrast, charcoal C is ~20% of total C in the mineral soils, irrespective of recent burn history and soil depth.

The contribution of charcoal C to the total carbon in mineral soils is in the middle of the range of published values for studies that used the peroxide-acid method. Charcoal C concentrations reported by Kurth et al. [12] range from 14 to 26% of total C in the upper 10 cm of Ponderosa pine forest mineral soils. Boreal Saskatchewan forest soils range from 18 to 22% of total C in the A horizon [28]. Forest soils in the Sierra Nevada Mountains contain charcoal C that represents 10–15% of total C in the upper 6 cm of the mineral soils [13].

Soil texture may explain the variation in charcoal contribution to total C in mineral soils. Higher percentages of silt plus clay correspond to a higher percentage of total C as charcoal C. For example, based on data from a Ponderosa pine forest [11,12,29], our evaluation indicates charcoal C as percentage of total C is positively correlated with silt plus clay concentration In the boreal transition zone of Saskatchewan, lacustrine soils with 85% silt plus clay have 22% of total C as charcoal C, while fluvial soils consisting of 41% silt plus clay have only 18% of total C as charcoal C [28]. Similarly, Sierra Nevada sandy loams and one silty loam contained 10–15% of total C as charcoal C [13], whereas the Siskiyou-LTEP loam and clay loams had 20% charcoal C. Soil organic C, in general, is stabilized by clay and silt [30,31,32]. The enhanced percentage of charcoal C with greater silt plus clay suggests charcoal is preferentially stabilized compared with other forms of organic C.

In the burned treatments, charcoal C in the O horizon averaged 35 kg C ha⁻¹ prior to fire and 311 kg C ha⁻¹ one year following fire (Figure 1). The increase in charcoal C was less for the thinned wildfire treatment than for the unthinned wildfire and thinned prescribed fire treatments (Figure 2). For comparison, the charcoal C in the O horizons of a montane Ponderosa pine forest was 10–336 kg C ha⁻¹ [7], and stands exposed to recent multiple fires contained three times more charcoal than stands that had not burned in greater than 79 years [33]. In boreal Swedish forest, somewhat greater values of O horizon charcoal C occur, ranging between 140 and 1620 kg C ha⁻¹ [7,20,34]. However, use of different methods to quantify charcoal C precludes absolute comparisons among the studies. As shown by Hammes et al. [35], the various charcoal C quantification methods isolate and measure an array of black C materials along the continuum. The Siskiyou LTEP values are based on physical separation followed by perioxide-oxidation digestion, whereas Ponderosa pine results are from physical separation and the assumption that 80% of charcoal mass is C [4] and boreal Swedish results are from nitric acid and benzenepolycarboxylic acid chemical digestions [20,35].

The charcoal C in the Siskiyou-LTEP mineral soils is the result of long-term, multi-century influences as well as recent disturbance. Charcoal C averaged 2093 (SE ± 107) kg C ha⁻¹ in pre-fire and 1904 (SE ± 101) kg C ha⁻¹ in post-fire surface (0–3 cm) mineral soil. Pre-fire charcoal C averaged 5217 (SE ± 469) kg C ha⁻¹ and post-fire averaged 5317 (SE ± 242) kg C ha⁻¹ in subsurface (3–15 cm) mineral soil (Figure 1). The higher charcoal C masses in the mineral soil compared to O horizons of the Siskiyou-LTEP study (Figure 1) agree with results from the Ponderosa pine forest, where mineral soil contained 10–30 times more charcoal than O horizon [12,34]. The mineral soils reflect an important pool of charcoal C influenced primarily by the long-term input of charcoal into the forest soils.

Comparing the Siskiyou-LTEP soils to published peroxide-acid digestion values from dry Ponderosa pine forest soils [11,12], boreal forest soils [28], and Sierra Nevada forest soils [13], we can identify variables that may influence the long-term charcoal C storage in mineral soils. The contribution of charcoal C to soil C pools may vary with fire history and fire intensity. Under a fire regime with short fire return interval, the frequent fires may completely burn charcoal deposited on the forest floor by previous fires before it can become mixed into the mineral soil. In contrast, infrequent fires may allow time for charcoal incorporation by mixing, as proposed by Brimmer [33] and DeLuca and Aplet [7]. Also, higher temperature fires have the potential to completely volatilize or convectively transport charcoal particles, leaving less charcoal on the forest floor than a lower temperature fire [7,34]. These processes would lead to a positive relation between mineral soil charcoal and fire return interval. Conversely, in a modeling study of thinned Ponderosa pine, the opposite trend was found, with more mineral soil charcoal resulting from a shorter fire return interval [7].

Comparison of the Siskiyou LTEP site with a dry Ponderosa pine forest supports the positive relationship between fire return interval and charcoal C in mineral soils. In a Ponderosa pine forest, the top 3 cm of mineral soil contains 450–780 kg charcoal C ha⁻¹ and the fire return interval is between 10 and 50 years [11]. The Siskiyou LTEP mineral soil, sampled the summer following the Biscuit Wildfire, has twice as much charcoal C (Figure 1) and more than twice the fire return interval. The site is situated on the west side of the Siskiyou Mountains where the fire return interval ranges from 90 to 150 years [36], which is much longer than further inland [37]. The lower fire return interval in the Ponderosa pine forest may have volatilized charcoal particles deposited from past fire events, leaving less charcoal C accumulation in mineral soils as suggested by DeLuca and Aplet [7].

Evaluation of charcoal in the forests of the Sierra Nevada Mountains [13] supports the positive relation between mineral soil charcoal and fire return interval. In that study, the top 6 cm of mineral soil contained 1000–2100 kg charcoal C ha⁻¹ in an oak woodland site, 2000–4500 kg charcoal C ha⁻¹ in a mixed conifer site, and 3500–4600 kg charcoal C ha⁻¹ in a red fir forest [13]. The fire return interval for the low-elevation woodland site is less than ten years, the mixed conifer site burns every 10 to 20 years, and the red fir site burns at 20 to 30 year intervals [13]. However, the Siskiyou-LTEP results do not extend the positive relation between mineral soil charcoal and fire return interval. The Siskiyou-LTEP charcoal C masses, 2600–3900 kg charcoal C ha⁻¹ for the top 6 cm of mineral soil, are similar to the mixed conifer and red fir sites of the Sierra Nevada soils, even though the Siskiyou fire return interval is much longer at 90–150 years [36].

Bélanger and Pinno [28] reported charcoal C for a boreal forest in Saskatchewan that range from 4000 to 11,000 kg charcoal C ha⁻¹ in the A horizons (depths of 7–13 cm) where the fire return interval is about 75 years or less where agricultural burning occurred, but 15 to 40 year intervals prior to 1890 [28]. Charcoal C masses from the Siskiyou-LTEP soils are almost three times greater than masses for Swedish and Siberian boreal forests, which developed under greater fire return intervals [7,20,34,36]. The Siskiyou-LTEP soils are consistent with the lower range of the boreal values and do not support the idea of a positive relationship between fire return interval and charcoal accumulation. The diverse results of these comparisons suggest that other environmental or inherent soil differences may play a large role in the processes involved in long-term charcoal C storage, such as soil texture, discussed above.

At the Siskiyou LTEP, change in surface mineral soil charcoal C did not differ among burn treatments (Figure 2), but taken together, the burn treatments decreased charcoal C by several hundred kg charcoal C ha⁻¹ compared with the unburned treatment (p = 0.06; unburned mean change = 175, SE = 211, n = 2; burned mean change = −310, SE = 101, n = 6). The loss of charcoal C corresponds to the loss of surface mineral soil via convection erosion during the fire or by post-fire erosion [21]. At the time of sampling, one year post-wildfire, soil may not have been exposed to enough mixing events, leaving the majority of the O horizon charcoal created by the 2002 Biscuit Wildfire remaining on the surface. Accumulation of charcoal in mineral soils is heavily dependent on biological and physical mechanisms that incorporate charcoal into lower soil horizons [7,38]. Other evaluations of recent wildfire influences on mineral soil charcoal C have used a chronosequence approach, which compares proximal sites that have been exposed to wildfire at different times in the past. These have produced varied results and may confound spatial differences with wildfire effects. For example, at a Ponderosa pine forest, mineral soil charcoal C mass was greater for a 12-year-old burn than 70-year-old burn [11], possibly indicating increase in charcoal by the recent wildfire. But at the same location, mineral soil charcoal C concentrations were similar for 12-year-old and 96-year-old burns [12], providing a different interpretation.

Charcoal production rates ranged from 1 to 8% of woody fuels consumed (Figure 3). The rate was negatively correlated (r = −0.99, p < 0.001) with amount of woody fuels consumed, resulting in the experimental unit with the greatest fuel loss having the lowest mass of charcoal produced and experiencing the highest fire temperatures, convection, and erosion (Figure 3). This supports the concept that higher fire severity can produce not only a lower percentage of charcoal, but also a lower absolute mass of charcoal. Different percentages of charcoal are produced under different fire conditions. In an experimental field study in Florida forest, prescribed fire moving with the wind produced less than half the charcoal percentage as fire moving against the wind [18]. Fire intensity, oxygen availability and fire duration were cited as possible causes of the difference.

The Siskiyou LTEP charcoal production rates fall within the range observed under a variety of fire and fuel conditions, substrates examined, and measurement approaches. In a prescribed fire in Florida scrub oak forest, pre and post-fire sampling indicated burned stem and leaf biomass yielded 4–6% charcoal from initial mass [17]. In another prescribed fire in Florida, wood experimentally placed on the forest floor produced 0.4–1.5% charcoal [18]. In a lodgepole pine (Pinus contorta) forest in a Yellowstone National Park wildfire, Wyoming, USA, post-fire analysis indicated 8% of coarse woody debris consumed by the fire was converted into charcoal [19]. In experimental fires, particle traps indicated 2.2% of the total fuel consumed was converted to charcoal in pine-spruce forest, Northwest Territories, Canada [28], and 2.0% was converted in pine forest, Siberia [38]. The general agreement among charcoal and charcoal C production rates is impressive, given the different techniques used for measurement, sampling and analysis, and their related uncertainties. For example, the Siskiyou LTEP rates do not account for very small (<0.5 mm) or large charcoal (>20 mm).

For assessing C budgets, Forbes et al. [39] emphasized distinguishing recalcitrant black C and expressing rates as black C produced as a percentage of total C consumed by fire, rather than charcoal produced as a percentage of biomass burned. Siskiyou LTEP O horizon charcoal particles averaged only 338 g charcoal C kg⁻¹ charcoal, where the charcoal C is defined as resistant to peroxide-acid oxidation. This translates to 1–5 % of total fuel C consumed by fire became charcoal C. This is consistent with Forbes et al. [39] summary that <3% of total C consumed in forest fires becomes black C.

The rate of charcoal production is a key consideration in modeling the influence of fire on soil C processes [40]. A number of laboratory and field studies identify the most important factors in assessing charcoal production rates. Fire intensity, fire return interval, vegetation type, fire behavior, and fuel loading represent common factors identified [5,7,13,16,18,38,40]. The Siskiyou-LTEP results provide a clear relation between fuel consumption and charcoal production under wildfire and prescribed fire conditions.

The Siskiyou LTEP site was established on the Rogue River—Siskiyou National Forest in 1992 in relatively homogeneous stands that had naturally regenerated after a wildfire in 1881 [21,40]. The study site is located about 25 km southeast of Gold Beach, OR at an elevation ranging from 750 to 900 m above sea level. Based on 1971 to 2000 values, mean January temperature is 4 °C, mean July temperature is 18 °C, of which only 10 cm falls between June and September [41]. Forest composition is dominated by Douglas-fir (Pseudotsuga menziesii var. menziesii (Mirb.) Franco) with knobcone pine (Pinus attenuata Lemmon) and sugar pine (Pinus lambertiana Dougl.). Hardwoods dominate the second story as tanoak (Lithiocarpus densiflorus (Hook. & Arn.) Rehd.), giant chinquapin (Chrysolepsis chrysophylla (Dougl. ex Hook.) Hjelmqvist var. chrysophylla), and madrone (Arbutus menziesii Pursh.) [21]. Soils are Typic Hapludults and Typic Dystrochrepts developed on a parent material from weathered sandstone and schist-phyllite [21,22,43]. As indicated by Bormann et al. [21] the mean C horizon depth is about 35 cm. The total soil inorganic material is comprised of 56% <2 mm, 7% 2–4 mm, and 37% >4 mm, and the <2 mm inorganic material is distributed as 26% clay, 37% silt, and 36% sand [21,22,42].

The original experimental design included 27 experimental units that encompassed thinning, clearcutting, and woody debris manipulation in 1997, and prescribed burning in 2001, as described in detail by Bormann et al. [21] and Homann et al. [22,43]. The Biscuit Wildfire and associated back-burn operation burned 13 of the units in 2002. Of relevance to this study are unthinned and thinned experimental units. The unthinned units were unmanaged and contained 80–110-year-old Douglas-fir (~1000 trees ha⁻¹), some knobcone pine and an understory of tanoak. The thinned units had a relative density averaging 240 trees ha⁻¹; all downed woody material >3 m in length and >5 cm in diameter at the small end was removed, except downed tree tops were retained on units that subsequently underwent prescribed burning. Prescribed burn treatments were fulfilled in the fall of 2001 with light surface fire and no fire reaching into the tree crowns [44]. For this soil charcoal study, two experimental units were evaluated for each of the following categories: unburned, unthinned wildfire, thinned wildfire, and thinned prescribed fire.

The temperatures, conifer mortality, and fire severity rating are displayed in Table 2. Melted aluminum tags denoting sampling points on the experimental units indicated the percentage of area that burned above 700 °C [21]. Conifer mortality was based on pre and post-fire measurements in five 18 by 18 m tree-measurement plots in each experimental unit [21]. BAER burn severity ratings correspond to a measure of soil impact.

Soil samples for this study were obtained from the LTEP soil archives, Corvallis, OR. As described by Homann et al. [22,43] and Bormann et al. [21], soil samples were collected at up to 16 points in each experimental unit before and after fire. Sampling before fire occurred between July and October in 1992 on units ICC and IL1 and in 1995 for the other units. Sampling after the fires occurred in 2003 on all units. Within each experimental unit is a central 2 ha measurement area surrounded by a 4 ha buffer. Measurement areas are further designated by a 25 m by 25 m permanent grid system. Sampling points were proximal to the grid intersections. The measurements from 2003 were ~3 m from 1992 and 1995 points. Where obstructions, usually in the form of trees, rocks, logs, or roots, prohibited sampling access, two additional attempts were made 1 m from the sampling point.

Soil samples were taken perpendicular to the slope, and slope angle was measured. At each point, the O horizon was collected from within a 30 cm diameter ring in 1992 and 1995. After removal of the O horizon, the mineral soil was sampled with a 10 × 15 cm soil corer to a depth of 30 cm. The mineral soil from the corer was extracted and separated into layers. The A horizon was sampled to a varied depth of up to 5.0 cm (average of 3.6 cm), and the B1 horizon was from the bottom of the A horizon to 15 cm. In 2003, similar procedures were used, except the O horizon was collected from within a 21 cm diameter ring and the A horizon was collected to a constant depth of 3 cm from within a 10 × 15 cm frame prior to coring. The separate collection of the A horizon was to prevent the fragile soil structure of the A horizon from being compromised during coring. The O horizon samples were hand-sorted into >4 mm rocks, wood >6.4 mm diameter, >4 mm charcoal, and remaining soil material. One sampling point on an unburned unit was eliminated from analysis because it had >4 mm charcoal from unknown source that was an order of magnitude greater than the other unburned and pre-fire sampling points. Mineral soil samples were sieved into the <4 mm soil fraction and >4 mm fraction, which contained rocks, roots, and large charcoal. Then the fractions were oven-dried at 70 °C and weighed. Slope-corrected soil masses (kg ha⁻¹) were calculated for each horizon. Within each experimental unit, a composite of the soil fraction was made for each horizon. This composite was mass-weighted from the individual sampling points.

Charcoal particles were separated from the O horizon soil composites following methods similar to Brimmer [33]. The O horizon composites were split into subsamples if the entire sample mass was more than 200 g, filtered through a 0.5 mm sieve if <0.5 mm particles were abundant, then spread out on a white tray to be hand separated with the aid of supplemental light. O horizon size fractions less than 0.5 mm were not considered due to the difficulty in visual determination. For 1995 ACC, ATB, and ATN units and 2003 C, BTB, and BTN units, the charcoal was separated into <4 mm and 4–20 mm fractions. For 2003 ACC, ATB, ATN, BCC, BTB, and BTN units, 4−20 mm charcoal particles were hand-separated from A and B1 horizons from individual sampling points. Charcoal particles >20 mm in size were rare and were not quantified in this study.

Charcoal C was measured in the following samples: surface (A) and subsurface (B1) soil horizon composites, 0.5–20 mm charcoal particles from selected O horizon samples, and charcoal and soils from the black C reference materials established by Hammes et al. [35]. Charcoal C was quantified by the method of Kurth et al. [12]. Charcoal and soils were dried for 24 hr at 60 °C and then exposed to the peroxide-acid digestion, as follows. One gram of mineral soil sample or 0.5 g charcoal was ground to a fine powder with a ball mill and placed into a 125 ml Erlenmeyer flask in which 20 ml of 30% H₂O₂ and 10 ml of 1 M HNO₃ were added and occasionally swirled at room temperature over a 30 min period. The flask was then heated in a water bath to 90 °C for 16 hours. After digestion, the suspensions were filtered through Whatman #2 filter papers (Whatman Ltd., Lawrence, KS, USA); filter papers were dried at 60 °C for 24 hours and weighed. Papers had been weighed prior to filtration and mass of residual material was determined as the difference in mass before and after filtration. Residual material was ground with mortar and pestle. Digestions were replicated on two different days.

Total C concentrations prior to and after digestion were determined on a Thermo NC 1112 Analyzer (CE Elantech, Inc., Lakewood, NJ, USA). Siskiyou-LTEP soils were measured against standard curves created with atropine (70.6% C). Total C of the charcoals and soils before digestions were assumed to represent organic C; Siskiyou soil pH is <5, indicating no or limited inorganic C. Total C after digestion is reported as charcoal C and assumes that all non-charcoal organic C is consumed during the peroxide-acid digestion. Duplicate measurements of charcoal C had an average coefficient-of-variation of 17%.

To allow future cross-referencing with other studies, black C reference chars and soils were acquired and exposed to the peroxide-acid digestion method [35,46,47]. The chars had been formed from grass straw (Oryza sativa) and chestnut wood (Castanea sativa) that had been collected in southern Switzerland. The chestnut wood was debarked and grass bundled then charred at 450 °C for 5 hours and ground when cooled. The H/C, O/C ratios, organic carbon content, and other chemical and physical characteristics of the chars are reported in Hammes et al. [46,47]. The Australian Vertisol is a sandy clay soil sampled at 0–10 cm depth at Urrbrae, South Australia, containing 37% clay [35]. The German Chernozem soil is a light sandy clay sampled at 20–60 cm depth in the Hildesheim-Braunschweig region, Harsum, Germany, and consists of 19% clay [35]. Our peroxide-acid charcoal C concentration, as % of organic C, was 9.1% for the Chernozem, 17.0% for the Vertisol, 19.0% for grass char, and no charcoal C was measured for the wood char.

Charcoal C concentrations of mineral soil samples and charcoal particles were determined with the following calculation:

    (1)

where C_digested is the C concentration in the digested sample and M_digested is the mass of the digested sample relative to the undigested sample.

Charcoal C concentrations for O horizon soils were calculated with the following formula:

    (2)

where C_particle is the concentration of charcoal C in charcoal particles from Equation (1); and M_particle is the mass of charcoal particles in the O horizon; C_particle was measured on a subset of samples (mean 338 g charcoal C ha⁻¹, SE ± 51, n = 2) and applied to all O horizons.

O horizon charcoal C mass (kg C ha⁻¹) was calculated by multiplying charcoal C concentration by O horizon soil areal mass. Mineral soil charcoal C mass for post-fire samples was calculated by multiplying charcoal C concentration by <4-mm soil areal mass for each of the surface (0–3 cm) and subsurface (3–15 cm) layers. To compare post-fire mineral soil C mass with pre-fire values, pre-fire mineral soil C mass for each layer was determined by the comparable layer approach, which accounts for bulk density changes and mineral-soil loss resulting from fire, as detailed in Bormann et al. [21] and Homann et al. [22].

The Biscuit Wildfire disrupted the original LTEP experimental design, but replicated experimental units (n = 2 for each of the four treatments: unthinned wildfire, thinned wildfire, thinned prescribed fire, and unburned) allow examining the change in charcoal C mass from before to after fire. For each of the O horizon, surface (0–3 cm) mineral soil, and subsurface (3–15 cm) mineral soil layers, a single factor ANOVA followed by Tukey’s multiple comparison test was used to compare among the four treatments [48]. Pearson correlation was used to examine the relation between charcoal production and woody fuels consumed.