The Influence of Slash Management Practices on Water and Nutrient Dynamics in Longleaf Pine Forests

Abstract

(1) Silvicultural applications that manipulate woody debris loading and the structural composition of a forest can have both short and long-term effects on biogeochemical cycling. Longleaf pine forests have been the historically dominant community types throughout much of the Southeastern United States. Fire exclusion, hardwood encroachment, and resource exploitation have severely reduced the amount of remaining longleaf pine habitats, making ecological restoration necessary. The silvicultural treatments used to reestablish these communities have been widespread, leading to some skepticism regarding the sustainability of certain restoration practices. (2) This study aimed to understand how overstory manipulation and woody debris management affected soil water retention rates and nutrient availability. Using a randomized complete block design, abiotic responses to biomass harvesting, conventional harvesting, and mastication treatments were measured across a soil moisture gradient in the South Carolina sandhills. (3) Our findings indicate that mastication increased soil moisture retention rates by 37% and 41%, on average, compared to conventional harvesting and biomass harvesting, respectively. (4) Additionally, soil nutrient stocks did not decline following any management practice, indicating that both biomass harvesting and mastication treatments may not necessarily impact site productivity in a negative manner. These findings imply that mastication treatments keep moisture retention high and do not immediately change soil nutrient availability in longleaf pine forests. Long-term vegetation response studies should continue to document successional trends in conjunction with moisture retention rates and long-term nutrient pulsing.

1. Introduction

Developed over various landscapes in the Southeastern United States (e.g., flatwoods, sandhills, and clayhills), longleaf pine (Pinus palustris Mill.) forests are often classified based on soil moisture characteristics [1,2,3]. They are one of the most species-rich and diverse forest types outside of the tropics but are also one of the most endangered in North America, represented by only a small fraction of their historical range [4]. Longleaf pine trees thrive in well-drained sandy soils and are both fire-adapted and shade-intolerant, meaning that fire exclusion promotes successional development and inhibits reproduction [5,6]. Historically, these forests have been maintained by frequent (1–3 years) low-intensity fires that exploit the complex herbaceous layer to create a recurrent disturbance regime [7,8,9,10,11,12,13]. As a result, understory vegetation is frequently restored in conjunction with longleaf pines, as it is critical to seedling survival [14,15]. However, plant establishment and restoration are often limited by light availability when there is a densely established overstory, so treatments that decrease the basal area and create canopy gaps will increase light penetration and encourage the growth of early successional grasses and forbs [16].

Many longleaf pine restoration efforts focus on transitioning loblolly pine (Pinus taeda L.), slash pine (Pinus elliottii Engelm), and hardwood stands to longleaf pine forests, using a combination of prescribed burning, herbicide, and mechanical treatments [15,17,18,19,20,21]. Conventional harvesting applications include significantly reducing the basal area, planting longleaf pine seedlings, and employing release treatments, which often consist of 1–2 herbicide applications followed by prescribed burning [15,22,23]. However, establishing herbaceous vegetation in sites such as the South Carolina sandhills can be challenging, as they are often highly eroded and experience water and nutrient deficiencies [24,25,26,27,28]. For example, Van Eerden (1997) found that bunchgrass seedling establishment was lower on xeric sandy sites compared to sites with higher quantities of loam and silt, suggesting that moisture retention was a limiting factor in this region [25]. As a result, restoration methods that increase moisture availability and maintain nutrient stocks could aid in establishing understory vegetation.

With new market demands, alternative silvicultural treatments, such as biomass harvesting and mastication, are sometimes used in restoration, but concerns about their applicability and long-term sustainability have arisen with increased use [29,30,31,32,33,34,35,36,37]. The removal of harvesting residue (i.e., logging slash) and alterations in the spatial distribution of masticated debris influence soil hydrology and nutrient dynamics, but the effects have been widespread [38,39,40,41,42,43,44,45,46,47]. In sandy soils with low cation exchange capacity, soil nutrient capital directly correlates to the amount of soil organic matter present, which is subsequently affected by the quality, quantity, and distribution of input materials [33,36]. Removing tree components with higher nutrient contents than wood (e.g., leaves, cambium, and root tips) could potentially cause declines in long-term site productivity, but could also provide greater germination success [33]. Conversely, the deposition and configuration of masticated wood can result in short-term periods of nitrogen immobilization, directly influencing plant growth and species composition [43,46,47,48]. However, other studies have found that mastication increases inorganic nitrogen and moisture retention rates, which may stimulate the vegetation response [39,40,45,46,47].

By quantifying the abiotic environmental conditions following varying intensities of woody debris manipulation, we hope to identify the optimal restoration methods, while simultaneously setting the stage for future vegetation response studies within the region. Our specific objectives were as follows: (1) to quantify the changes to woody debris loading following different silvicultural treatments (i.e., conventional harvesting, biomass harvesting, mechanical mastication) that result in various spatial configurations and woody debris quantities, (2) to determine how manipulating woody debris during restoration timber harvest affects soil moisture retention rates and nutrient availability along a soil moisture gradient. We hypothesize that (1) the intensified removal of woody debris during the biomass harvesting treatment will result in a parallel decline in nutrient availability, (2) masticated debris beds will yield greater moisture retention rates that will be most apparent in xeric sites during periods of drought and (3) the addition of masticated woody debris will result in a short period of nitrogen immobilization.

2. Materials and Methods

The Hardscramble tract is 305 has and is located in the city of Camden in Kershaw County, SC, USA (coordinates: 34.26732, −80.65668) and exhibits a humid, subtropical climate, with year-round rainfall and long hot summers. Precipitation averages about 1100 mm annually, with average minimum and maximum temperatures ranging from 10.3 °C to 24 °C [49]. The underlying geology of the site is comprised of Cretaceous sediments of the Middendorf Formation, which, in Kershaw county, range from 0 to 107 m in depth and sit on top of crystalline rocks from the Paleozoic age [50]. Thick beds of Cretaceous clays form the first visible layer and are located directly beneath deep layers of highly permeable quartz sand [27]. The property retains fourteen different soil series, all of which classify as either Ultisols or Entisols, and is characterized by various habitats, ranging from xeric mature longleaf pine stands to mesic bottomland hardwood stands. Within the property, the study area is a 28.7 ha subsection in the northeast corner (Figure 1). While mature longleaf pine was present, the majority of the study area was classified as an uneven-aged mixed pine and oak forest, with a loblolly overstory and a densely stratified midstory of hardwood species. Intense shading, caused by a prolonged lack of forest management, inhibited understory development and longleaf pine regeneration for much of the study area.

Pre-treatment data were collected from August 2019 to August 2020, and post-treatment data were collected from May 2021 to September 2021. In Fall 2019, a landscape ecological classification (LEC) system, developed for the Hilly Coastal Plain province of South Carolina, was used to classify sites along a soil moisture gradient, with the following four classifications ranging from the wettest sites to the driest sites: mesic, submesic, subxeric, xeric [3]. This designation was used to create a randomized complete block design that organized the 28.7 ha into four distinct soil moisture blocks. Each soil moisture block was subdivided into three compartments, where we implemented our treatments. Three random sampling points were placed within each treatment using ArcGIS mapping applications (36 total sampling points) (Figure 2a). A seed-tree timber harvest was conducted from November to December of 2020 over the entire study area, using mechanized harvesting equipment (i.e., feller buncher, grapple skidder and knuckle-boom loader). Post-harvest woody debris manipulation simulated the site conditions of the following three different silvicultural treatments: conventional harvesting, biomass harvesting, and fuel mastication. Harvesting slash was scattered across the site for conventional harvesting treatments, removed for biomass harvesting treatments, and masticated for the remaining treatments.

Prior to the implementation of our silvicultural treatments, woody debris measurements and water retention rates were collected at each point using the line intersect method and a series of tensiometers (Soil Measurement Systems, Tucson, AZ, USA) placed at each location [51,52]. Thirty-six 15.24 cm (6 in) tensiometers were used to measure biweekly soil water potential (kPa) from 20 July 2020 to 27 August 2020. Woody debris, with diameters larger than 7.62 cm, were labeled as coarse woody debris (CWD) and volumes were calculated using the conic paraboloid equation (A_a = cross-sectional area at the upper end; A_b = the cross-sectional area at the lower end; L = length) [53].

Conic-paraboloid volume = (L/12) × (5A_a + 5A_b + (A_aA_b)^1/2)

(1)

Woody debris, with diameters less than 7.62 cm, were labeled as fine woody debris (FWD) and subdivided into three categories based on diameter size (size class 1 = 0–0.64 cm, size class 2 = 0.64–2.54 cm, and size class 3 = 2.54–7.62 cm). FWD volume was estimated using Huber’s volume formula (A_m = cross-sectional area at the midpoint; L = length) [53].

Huber’s volume = L × A_m

(2)

Estimated per unit values were found using Waddell’s (2002) equations based on DeVries’ (1973) formulas (L= total length of transect line; V_m = volume in cubic meters of individual piece; l_i = length of individual piece; f = 10,000 m²/ha; SpG = specific gravity of wood; DCR = decay class reduction factor) [51,54,55].

Cubic meters per ha = (π/2 L) × (V_m/l_i) × f

(3)

Kilograms per ha = (cubic meters per ha) × 1000 kg/m³ × SpG × DCR

(4)

For FWD biomass estimates, volumes were multiplied by the average bulk density for Quercus spp. (579.7 kg m⁻³) and an assumed decay class reduction factor of 0.80 [56]. Average oak bulk density was used to avoid underestimation of FWD, which often happens when using average pine bulk densities, especially in mixed stands [56]. Carbon estimates were found by multiplying the per unit biomass weight by a conversion factor (0.491 for hardwoods and 0.521 for softwoods) [54].

Post-treatment, each tensiometer was replaced and measured weekly from May 2021 to September 2021. A 4 m × 20 m strip plot was positioned approximately 2.5 m from the sampling point at the same azimuth as the pre-harvest woody debris transects [38]. Within the post-treatment strip plot, four destructive FWD samples were taken at each plot corner, four soil samples were taken at the midpoints of each plot boundary line, and three bulk density samples were taken at 5 m intervals in the center of each CWD plot (Figure 2b).

The length, large end diameter, and small end diameter for CWD logs were recorded if the length was larger than or equal to 1 m, if the large end diameter was greater than or equal to 15 cm, and if the small end diameter was greater than or equal to 7.62 cm [54,57]. Litter and duff were combined and FWD was separated into diameter size classes. All samples were washed, air-dried for 48-h before being weighed, dried in an air-dry oven at 105 degrees C for 72 h, and finally reweighed. Dry weights were converted to per-unit values (kg ha⁻¹) using sampling estimations and bulk densities provided by the Forest Inventory and Analysis (FIA) program [56]. Per-unit values (kg ha⁻¹) were obtained by scaling up our strip plots, using the combined weight of the woody debris inventoried within the boundaries of each plot.

All the soil samples were sandy and coarse-textured, with a clay layer (B-horizon) at a depth of at least 101.6 cm. The soils were tested for pH, buffer pH, extractable elements (P, K, Ca, Mg, Zn, Mn, Cu, B, Na, S), organic matter, nitrate-nitrogen (${\mathrm{NO}}_3^{-}\mathrm{N}$), and soluble salts. Calculations for the cation exchange capacity (CEC), acidity, and percent base saturation were also included in the analysis. Soil sample concentrations were classified as either sufficient or insufficient based on the Clemson University Agricultural Service Laboratory Soil Test Rating System, which defines soil fertility status based on a nutrient element concentration range [58]. In order to supplement our soil sample findings, weekly soil solution samples were obtained from twelve 30 cm porous cup suction lysimeters (Hanna Instruments HI89300-30) from June 2021 to August 2021. One lysimeter was placed 15.24 cm into the soil in each treatment. Using spectrophotometry, ammonium, nitrate, and phosphate concentrations were measured using the Environmental Protection Agency (EPA) compliant FIA-012 method, the FIA-026 cadmium reduction method (EPA 353.2, SM 4500-NO3), and the FIA-073 sequential flow injection method (EPA 365.1), respectively [59].

All data transformations and analyses were conducted using Microsoft Excel [60] and R [61]. The treatment and moisture block variables were converted into factor data, measures of central tendency and dispersion for our dependent variables were determined, and finally we tested for normal distribution using normal Q-Q plots and the Shapiro–Wilk test on the ANOVA residuals. Levene’s test was used to measure the homogeneity of variances. Data that were not normally distributed were log-transformed to meet the assumptions of normality. In order to better understand the effects of our treatments on soil moisture characteristics and nutrient availability, we used the data compiled from 2020 to 2021 to investigate the different trends. Soil nutrient availability and downed woody debris were analyzed using a two-way analysis of variance (ANOVA), and time-integrated measures (matric water potential) were analyzed with a repeated-measures ANOVA using a Bonferroni correction. Tukey’s HSD post-hoc tests were performed for pairwise comparisons and, for any significant interactions, we developed an interaction plot using least-squares means tests for different factor combinations.

3. Results

3.1. Woody Debris

A preharvest inventory of CWD found that there were no significant differences in the mean amounts of biomass (kg ha⁻¹) or the CWD predicted weight of carbon (kg ha⁻¹) across the soil moisture blocks or treatments. There was an average of 168.3 kg ha⁻¹ (min = 11.6 kg ha⁻¹; max = 3303.9 kg ha⁻¹) across the entire study area, with only two samples indicating more than 1000 kg ha⁻¹. The post-treatment CWD analysis found that there were significant differences in the means between treatments for CWD biomass (p = 0.030) and CWD carbon estimates (p = 0.028) (Figure 3). However, for both woody biomass and carbon, there were no significant findings between soil moisture blocks (p = 0.188; p = 0.284, respectively) and there were no interaction effects between the independent variables (p = 0.056; p = 0.052, respectively). The amount of CWD biomass increased across the treatments from our masticated sites to our conventional harvest sites (Figure 3). The pairwise comparisons found that, on average, conventional harvesting produced 7349 kg ha⁻¹ (142%) more CWD and 3645.8 kg ha⁻¹ (141%) more CWD carbon compared to our mastication treatment (Figure 3).

The average pre-treatment FWD across all sites was 2802.4 kg ha⁻¹ and there were no significant mean differences across the soil moisture blocks (p = 0.080) or across treatments (p = 0.775). Post-treatment destructive sampling found that the mean dry weight for litter and duff was 4046.7 kg ha⁻¹ (SD = 2432.6 kg ha⁻¹) across the entire study area and that the means across treatments (p = 0.541) and soil moisture blocks (p = 0.535) were not significantly different (Figure 4A). Post-treatment duff and litter depths were not measured. However, using average pre-treatment duff and litter depths, we found that post-treatment bulk density averaged about 16.2 kg m⁻³.

Post-treatment measurements found significant differences between our treatments for size classes 1, 2, and 3 (p = <0.05; p = 0.014; p = 0.009, respectively) (Figure 4). On average, the mastication treatment produced 6642.4 kg ha⁻¹ and 6116.8 kg ha⁻¹ more woody debris in size class 1 and 7316.9 kg ha⁻¹ and 7236.0 kg ha⁻¹ in size class 2, compared to the biomass and conventional harvest treatments, respectively (Figure 4B,C). This trend reverses with our largest FWD size class, producing the least amount of woody debris per unit area in masticated treatments (Figure 4D). The conventional harvest treatment produced the largest amount of woody debris in size class 3, and significantly more than the mastication treatment (75% more kg ha⁻¹) (Figure 4D). For all moisture blocks and treatments, the harvest generated an average of 5328.5 kg ha⁻¹, 11,978.5 kg ha⁻¹, and 16,398.5 kg ha⁻¹ for our size classes in ascending order.

3.2. Soil Moisture Availability

The pre-harvest measures of soil water matric potential yielded significant differences in the means across dates (p = <0.05), but not across moisture gradients (p = 0.438) or treatments (p = 0.177). Following the implementation of our treatments, we found significant differences between soil moisture blocks (p = <0.05), treatments (p = <0.05), and dates (p = <0.05), indicating that, in addition to sporadic rainfall events, our treatments influenced soil water availability for vegetation (Figure 5). Our statistical analysis produced significant interaction effects between our moisture gradient and our treatments (p = <0.05), as well as between our treatments and recording dates (p = <0.05) (Figure 5). Our masticated treatment remained consistently high in water availability across dates, and the greatest differences between treatments were observed in our xeric sites during dry periods (Figure 5). Soil tension increased along our moisture gradient from mesic to xeric, but did not differ significantly between biomass and conventional harvesting treatments. However, for five sampling periods during the growing season, moisture became a limiting (<−30 kPa) factor or near-limiting factor for the biomass and conventional harvest treatment, while soil water in the masticated treatments never dropped below −20 kPa (Figure 5). For these five sampling dates, soil moisture was significantly higher in the masticated treatments compared to the other treatments (Figure 5).

3.3. Soil Quality

Across all treatments, soil pH was strongly acidic (M = 4.9, SD = 0.2) and did not exhibit a significant difference in the means across treatments (p = 0.607) or between soil moisture blocks (p = 0.338) (

Table 1. Soil composition derived from soil samples taken from the Hardscramble property longleaf pine restoration site in Camden, SC, USA. For each soil moisture block and treatment, soil pH, extractable elements (P, K, Ca, Mg, Zn, Mn, Cu, B, Na, S), cation exchange capacity (CEC), nitrate-nitrogen (${\mathrm{NO}}_3^{-}\mathrm{N}$), and organic matter were measured.

Moisture Block	Treatment	Soil pH	P (kg/ha)	K (kg/ha)	Ca (kg/ha)	Mg (kg/ha)	Zn (kg/ha)	Mn (kg/ha)	Cu (kg/ha)	B (kg/ha)	Na (kg/ha)	S (kg/ha)	CEC (mmhos/cm)	${\mathbf{NO}}_3^{-}\mathbf{N}\mathbf{ppm}$	OM %
Mesic	Conventional	5	11.2	28.0	89.7	12.3	1.2	13.5	0.2	0.2	5.6	7.8	0.03	0	1.12
		4.9	3.4	43.7	100.9	28.0	0.7	11.2	0.1	0.3	6.7	12.3	0.03	0	1.83
		5	3.4	61.6	244.3	35.9	1.5	22.4	0.2	0.3	11.2	11.2	0.05	1	2.28
	Biomass	5	15.7	26.9	96.4	12.3	0.7	10.1	0.2	0.2	4.5	7.8	0.03	1	1.96
		4.9	3.4	30.3	68.4	15.7	0.8	7.8	0.1	0.2	6.7	13.5	0.05	0	1.39
		5.2	7.8	96.4	358.7	53.8	1.9	42.6	0.4	0.4	10.1	12.3	0.05	0	2.98
	Mastication	5.2	4.5	42.6	107.6	19.1	0.4	15.7	0.2	0.3	6.7	9.0	0.04	1	2.17
		5.2	6.7	40.4	97.5	23.5	0.9	44.8	0.3	0.3	9.0	9.0	0.03	0	1.74
		4.5	13.5	42.6	154.7	26.9	1.6	9.0	0.2	0.2	5.6	7.8	0.05	0	2.84
Submesic	Conventional	5	13.5	19.1	103.1	10.1	0.8	15.7	0.3	0.2	4.5	9.0	0.03	0	1.26
		4.7	17.9	44.8	130.0	23.5	1.3	19.1	0.3	0.3	7.8	9.0	0.06	0	1.52
		4.9	5.6	37.0	60.5	24.7	0.6	5.6	0.3	0.3	9.0	38.1	0.04	0	1.78
	Biomass	4.6	11.2	26.9	70.6	12.3	0.7	5.6	0.2	0.2	7.8	6.7	0.05	0	1.21
		4.8	39.2	37.0	49.3	10.1	0.8	6.7	0.2	0.2	5.6	6.7	0.05	0	1.19
		4.8	20.2	29.1	94.2	14.6	0.8	14.6	0.2	0.2	6.7	11.2	0.04	0	1.2
	Mastication	5.1	5.6	57.2	191.7	29.1	1.6	10.1	0.1	0.3	5.6	12.3	0.04	0	2.01
		5	11.2	67.3	233.1	33.6	2.5	26.9	0.3	0.3	9.0	10.1	0.05	0	1.79
		4.8	4.5	42.6	178.2	26.9	1.2	24.7	0.1	0.2	4.5	9.0	0.03	1	2.38
Subxeric	Conventional	4.8	2.2	16.8	34.7	7.8	0.7	5.6	0.2	0.2	5.6	14.6	0.03	0	1.66
		4.7	3.4	23.5	57.2	16.8	0.8	7.8	0.2	0.4	13.5	14.6	0.04	0	1.68
		4.9	3.4	41.5	146.8	31.4	1.1	15.7	0.3	0.3	13.5	14.6	0.04	0	1.63
	Biomass	4.6	5.6	91.9	520.1	78.5	3.0	52.7	0.3	0.4	17.9	11.2	0.07	0	3.89
		4.8	5.6	81.8	378.8	74.0	2.6	46.0	0.3	0.4	17.9	13.5	0.07	0	2.92
		4.7	5.6	44.8	134.5	19.1	1.3	28.0	0.3	0.3	6.7	14.6	0.09	5	2.39
	Mastication	5.3	5.6	50.4	274.6	48.2	1.0	32.5	0.2	0.3	10.1	11.2	0.04	0	2.17
		5	5.6	47.1	182.7	32.5	1.5	42.6	0.2	0.3	11.2	10.1	0.03	0	1.40
		4.6	4.5	66.1	130.0	26.9	0.9	13.5	0.2	0.3	7.8	9.0	0.05	0	2.15
Xeric	Conventional	4.9	3.4	49.3	68.4	19.1	0.6	6.7	0.2	0.4	11.2	15.7	0.05	1	1.35
		4.7	5.6	60.5	171.5	31.4	1.8	21.3	0.2	0.3	5.6	11.2	0.05	0	2.58
		4.8	3.4	43.7	108.7	25.8	1.0	12.3	0.2	0.2	6.7	9.0	0.04	0	1.88
	Biomass	4.9	22.4	58.3	288.1	47.1	1.7	26.9	0.3	0.3	5.6	9.0	0.04	0	1.7
		5	17.9	33.6	95.3	16.8	0.9	31.4	0.4	0.3	6.7	9.0	0.03	0	1.32
		5.1	7.8	30.3	156.9	22.4	0.9	26.9	0.2	0.3	5.6	12.3	0.03	0	1.34
	Mastication	4.6	4.5	32.5	80.7	14.6	0.9	14.6	0.3	0.2	7.8	10.1	0.05	0	2.5
		4.8	12.3	48.2	193.9	25.8	1.1	51.6	0.3	0.3	11.2	14.6	0.04	0	1.92
		5.1	7.8	81.8	404.6	78.5	2.0	58.3	0.3	0.4	10.1	10.1	0.06	1	2.05

). There was a similar trend for the amount of extractable potassium (K), where low K levels were present across the study area (M = 20.8 ppm, SD = 8.7 ppm) and did not show significant differences between the treatments or soil moisture blocks (p = 0.234; p = 0.593, respectively) (Table 1). Nitrate nitrogen (${\mathrm{NO}}_3^{-}\mathrm{N}$) was low across all sites, with 80.6% of all samples having 0 ppm, 16.7% having 1 ppm, and only one sample showing 5 ppm (Table 1). The only elements that were sufficiently available were manganese (Mn) in 25% of the samples and magnesium (Mg) in 5.6% of the samples (Table 1). Extractable phosphorus (P) was available in insufficient amounts as well (M = 4.3 ppm, SD = 3.3 ppm), exhibiting low amounts of P (0–15 ppm) in 97% of our samples (Table 1). However, there was a significant difference in the means between treatments (p = 0.005) and soil moisture blocks (p = 0.005), with our submesic biomass harvest exhibiting the greatest amount of P per unit area. Pairwise comparisons found that, on average, P levels were 113% greater in biomass harvest treatments compared to our conventional treatments, and 210% higher in submesic sites compared to subxeric sites. However, there were no significant interaction effects (p = 0.248).

Lysimeter readings found that there were significant nitrate (${\mathrm{NO}}_3^{-}$) concentration differences across soil moisture blocks (p = 0.002), with pairwise comparisons showing higher average concentrations for the subxeric sites. The highest maximum concentration was 6.3 ppm and was located in a subxeric moisture block (

Table 2. Mean parameters of soil solution variable nutrient concentrations (ppm) at a 15.24 cm soil depth between woody debris manipulation treatments and soil moisture blocks in Camden, SC, USA. Parenthetical values are standard deviation.

Moisture Group	Treatment	$\mathbf{N}{\mathbf{O}}_3^{-}$			$\mathbf{N}{\mathbf{H}}_4^{+}$			$\mathbf{P}{\mathbf{O}}_4^{3-}$
		Predicted Peak	Average	Max	Predicted Peak	Average	Max	Predicted Peak	Average	Max
Mesic	Mastication	0.56 (0.03)	0.43 (0.17)	0.58	0.21 (0.34)	0.17 (0.24)	0.61	0.24 (0.16)	1.11 (1.84)	6.35
	Biomass	0.56 (0.01)	0.52 (0.10)	0.62	0.03 (0.02)	0.04 (0.06)	0.19	1.14 (1.57)	5.51 (13.40)	44.94
	Conventional	0.74 (0.05)	0.65 (0.12)	0.79	0.19 (0.18)	0.29 (0.18)	0.66	0.23 (0.04)	0.57 (1.06)	2.83
Submesic	Mastication	0.91 (0.35)	0.87 (0.40)	1.27	0.04 (0.02)	0.05 (0.03)	0.13	0.63 (0.75)	3.68 (9.75)	33.55
	Biomass	0.67 (0.19)	0.59 (0.18)	0.89	0.01 (0.01)	0.01 (0.04)	0.14	0.38 (0.35)	1.20 (2.84)	10.01
	Conventional	0.65 (0.13)	0.52 (0.18)	0.8	0.01 (0.01)	0.03 (0.05)	0.16	0.17 (0.06)	0.33 (0.36)	1.34
Subxeric	Mastication	3.35 (2.88)	3.33 (2.59)	6.32	0.01 (0.01)	0.03 (0.06)	0.19	0.26 (0.14)	0.77 (1.97)	5.01
	Biomass	2.13 (1.23)	2.03 (1.15)	3.16	0.01 (0.01)	0.04 (0.07)	0.24	0.18 (0.08)	0.94 (2.12)	7.48
	Conventional	2.17 (2.56)	2.07 (2.35)	5.12	0.05 (0.04)	0.05 (0.04)	0.12	0.41 (0.46)	0.92 (2.36)	8.08
Xeric	Mastication	0.57 (0.02)	0.52 (0.11)	0.64	0.07 (0.08)	0.08 (0.1)	0.35	0.27 (0.21)	1.71 (4.47)	15.74
	Biomass	0.75 (0.32)	0.65 (0.31)	1.11	0.03 (0.04)	0.01 (0.04)	0.08	0.19 (0.08)	0.39 (1.50)	4.57

). However, there were no significant differences or interactions between treatments (p = 0.988; p = 0.317, respectively). Additionally, we did not find significant mean differences in the Gaussian peak predictions for our treatments (p = 0.769) or soil moisture blocks (p = 0.063).

Comparatively, mean ammonium (${\mathrm{NH}}_4^{+}$) concentrations were low but were significantly different between soil moisture blocks (p = 0.003) and treatments (p = <0.001). On average, our biomass treatment produced significantly lower ammonium concentrations compared to all other treatments, and our mesic site produced significantly higher ammonium concentrations than all other soil moisture blocks (Table 2). We also found significant interaction effects between our independent variables, with conventional harvest sites on mesic soil producing very high concentrations (p = 0.002). However, there were no significant findings for the Gaussian peak predictions across or between the independent variables.${\mathrm{PO}}_4^{3-}$ concentrations showed no significant differences between the means or between the mean Gaussian peak predictions across or between independent variables. Maximum concentrations ranged from 1.34 to 44.94 ppm and the highest predicted mean Gaussian peak prediction was 1.14 ppm, with a majority of the results having less than 0.5 ppm (Table 2).

Soil bulk density (g cm⁻³) increased, on average, as the soil moisture gradient became drier and sandier. Significant differences were found between soil moisture blocks (p = 0.001) but not between treatments (p = 0.900), and there were no significant interaction effects (p = 0.752). Xeric and subxeric soil bulk densities were 15.6% and 13.3% higher, respectively, compared to mesic sites. Soil porosity was lower in the xeric sites, ranging from 43 to 45%, on average, across treatments, suggesting stronger matric potentials. Soil organic matter averaged 1.92% of the overall soil composition by weight across all sites. There were no significant mean differences between treatments (p = 0.166) or soil moisture blocks (p = 0.066). However, there was a significant interaction effect between our independent variables (p = 0.012). Biomass harvesting in the subxeric moisture block exhibited the greatest amount of soil organic matter and significantly higher amounts of organic matter compared to conventional harvesting treatments in the submesic (+1.55%) and subxeric sites (+1.41%), and biomass harvesting treatments in the submesic (+1.87%) and xeric sites (1.61%).

4. Discussion

4.1. Downed Woody Debris

The pre-treatment measurements of downed woody debris found that CWD and FWD loadings were similar across the study area, confirming that our site conditions before the harvest were relatively uniform. Post-treatment FWD and CWD were, on average, different across treatments. Masticated treatments had significantly more woody debris in size classes 1 and 2 compared to the other treatments, which showed no difference. Biomass harvest treatments exhibited the lowest amount of downed woody debris. While masticated treatment should have produced the same amount of woody debris as the conventional treatment, the latter was estimated to produce over 2000 more kg ha⁻¹ than mastication. This finding could be a result of our sampling methods, which assumed a consistent shape for the CWD logs. Additionally, we used a plot-based method for measuring all woody debris, which is less consistent than a hybrid methodology [38]. Due to the irregular shape of masticated wood particles, it is recommended that quantification of woody debris should come in the form of a hybrid methodology, in which masticated material is measured using plot-based measurements and larger, coarse woody debris is calculated using the standard planar intercept method [38,62].

4.2. Soil Moisture Availability

Soil water matric potential was consistent across all sites for the pre-treatment findings. Our most significant fluctuations were observed between dates that did not experience precipitation events close to our observation period. Expectedly, the matted debris produced by our mastication treatment had higher and more consistent moisture retention rates even during dry conditions and this was the only treatment where water was never a limiting factor for plant growth. Kreye et al. (2012) found that fuelbeds composed of 0–0.64 cm and 0.64–2.54 cm particles increased water retention, producing response drying times for fuels that ranged between 40 and 69 h [39]. Masticated treatments have not been found to influence drying rates for individual particles when compared to non-fragmented particles of the same size class [39,42]. However, the compacted nature of the bed slows the overall drying rate, potentially resulting in positive vegetation responses and pulses of available nutrients, known as the Assart effect [42,63]. A major concern is that the slower overall drying rates, the decrease in bare soil, and the structural configuration of the woody debris will limit prescribed fire capabilities and nitrogen availability in the short-term [39,40,47,64].

Our differences were magnified in the xeric sites, showing an increase in matric water tension in conventional and biomass treatments, likely due to mineral soil exposure. Loose sandy soils can produce additional heat through reradiation, which subsequently increases the evapotranspiration of surface moisture [65]. Masticated fuelbeds can reduce the rate of surface water evaporation, stabilize soil temperature, and slow the rate of infiltration [42,44,66]. These effects can positively influence vegetation responses, but can also influence when and to what extent certain nutrients become available [44].

4.3. Soil Quality

Overall, soil composition was characteristic of South Carolina sandhill ecosystems, exhibiting high acidity with extremely well-drained sandy soils across the entire site. We found similar findings to Brendemuehl (1967) and Faust (1976), who noted that sandhills’ fall line soils average less than 2% organic matter, with a soil pH range from 4.0 to 5.6 [65,67]. Low pH values can cause deficiencies in soil nutrients, especially nitrogen and phosphorus. Nitrification is highly sensitive to changes in soil pH, litter composition, and soil water content, essentially ceasing at values less than 5.0 pH [68,69,70]. Additionally, inorganic phosphorus is frequently deficient at pH values below 5.5, and these effects are often exacerbated by a lack of organic matter [71].

Soil bulk density varies greatly by soil type and depth, with porous soils containing large amounts of organic matter, silt, or clay having lower bulk densities (0.5–1.6 g cm⁻³) compared to sandy soils with less total pore space (1.3–1.7 g cm⁻³). Deeper sandy soil layers can have even higher rates of compaction (> 2.0 g cm⁻³) [72,73]. For sands and loamy sands, such as those present in the South Carolina sandhills, the USDA-NRCS identifies ideal bulk densities for plant growth as <1.60 g cm⁻³, with bulk densities > 1.80 g cm⁻³ restricting root growth. High bulk densities could, thus, affect long-term vegetation responses in xeric sites by either restricting plant growth or slowing the rate of percolation and enhancing water retention capacities. We found no significant differences between treatments, although bulk density increased in the xeric block. Long-term monitoring should be conducted across the study area to quantify the effects of decomposition on bulk density, especially in masticated sites, where organic material may be more likely to become incorporated into the mineral soil over time.

Due to high soil temperatures, organic matter in this region oxidizes quickly and releases nutrients that are readily leached from the upper soil levels by frequent rainfall events [27,65,74]. Of the 17 essential nutrients and the 6 primary nutrients (C, H, O, N, P, K) required for plant growth, nitrogen (N), sulfur (S), and phosphorus (P) are highly influenced by the presence of organic matter and are often limited in sandhill communities [71,75]. Moreover, following disturbances, subsurface pools of N and P are key indicators for ecosystem recovery and nutrient availability [59]. Longleaf ecosystems in sandhill regions are typically nutrient deficient, and while restorative methods that remove soil carbon (e.g., prescribed burning and biomass harvesting) might result in short-term changes to nutrient availability, they may have little long-term effect on soil quality [76,77]. For example, after only one year following prescribed fire in a longleaf pine stand in the Florida sandhills, C and N pools recovered to 67% and 76% of pre-treatment levels [78].

Inorganic nitrogen (${\mathrm{NO}}_3^{-}$ and ${\mathrm{NH}}_4^{+}$) is the most commonly limiting nutrient in longleaf sandhills, due to the low litter quality (high C:N ratio) and high soil acidity. High C:N ratios result in ammonium immobilization, while nitrification is inhibited at low pH values [43,48]. Inorganic phosphorus (${\mathrm{H}}_2{\mathrm{PO}}_4^{-}$, ${\mathrm{HPO}}_4^{2-}$, and ${\mathrm{PO}}_4^{3-}$) is essential for plant growth, energy transfer (ADP and ATP), reproduction, photosynthesis, and several other plant functions, frequently showing deficiencies in soils with pH values below 5.5 or above 6.5 [71]. However, increases in soil organic matter have been found to reduce the strength of P adsorption and the maximum phosphate buffering capacity [79]. Similarly, inorganic sulfur (${\mathrm{SO}}_4^{2-}$) is also considerably influenced by soil organic matter and is most important in the formation of chlorophyll. Sulfur is often deficient in coarse sandy soils with good drainage [70,71,80].

We observed that ${\mathrm{NO}}_3^{-}$ concentrations were considerably higher than ${\mathrm{NH}}_4^{+}$ concentrations but were still low (<1 ppm) in most of our soil tests and solutions. This finding was interesting, considering other longleaf restoration studies found that ${\mathrm{NH}}_4^{+}$ can supplement longleaf growth during growth-limiting environmental periods that exhibit low ${\mathrm{NO}}_3^{-}$ concentrations [69]. While we did observe a post-harvest spike of ${\mathrm{NO}}_3^{-}$ in our subxeric sites, there were no discernable trends for nutrient availability across our treatments. Wilson et al. (2002) found that soil temperature increases were the principal influence in producing large pools of inorganic nitrogen in xeric sites. However, this does not explain why there was not a similar pulse of nitrogen compared to our subxeric results [70]. We also noted a parallel spike in soil organic matter in our subxeric site, but it did not seem to influence ${\mathrm{NH}}_4^{+}$ or ${\mathrm{PO}}_4^{3-}$. Considering these soils exhibit a low CEC with very little organic matter, nutrients may have leached below the depth of our lysimeters.

Without an observed period of immobilization in the mastication treatment or a heightened deficiency in the biomass treatment, it could be argued that the initial disturbance response to these treatments is negligible. Coates et al. (2010) found similar results, in which fuel reduction treatments had little to no effect on soil properties 2–4 years following their study in the Southern Appalachians [81]. In an already nutrient deficient environment, we found that both biomass harvesting and mastication are essentially equally sustainable with regard to soil quality. Without nutrient deficiencies, biomass harvesting exhibited similar soil characteristics compared to conventional harvesting methods. Conversely, mastication treatments showed similar trends to other studies, which noted that the addition of masticated materials did not result in short-term changes to nutrient availability [47,82,83]. However, the additional and consistent water retention rates produced by that treatment could improve the long-term vegetation responses. Water availability often constrains the establishment and performance of native understory plants in longleaf pine forests [24]. Thus, any treatment that enhances water availability during the growing season could be a viable option for managers seeking to restore the understory community.

Due to the fragmented and compacted nature of masticated woody debris, decomposition typically happens at faster rates than coarse woody debris, decreasing the period of nitrogen immobilization and resulting in faster mineralization [46,47]. In turn, this could potentially influence long-term nitrogen dynamics. For example, Rhoades et al. (2012) noted an initial decline in plant-available nitrogen following mastication, but then observed a rapid rate of mineralization, showing an elevated level of available nitrogen (+32%) three to five years following the treatment [46]. Young et al. (2013) found similar results and concluded that mastication could improve pine seedling growth by increasing the available nitrogen and moisture retention [47].

It is also possible that we are currently experiencing a period of immobilization compared to pre-harvest concentration levels and that, because the C:N ratio would increase similarly across all treatments just in different spatial arrangements, our initial findings do not exhibit any differences between treatments. This could lead to a delayed mineralization process that manifests itself several years following the initiation of our treatments [46]. Pre-treatment concentrations were not evaluated, but long-term studies on mineralization could still indicate significant fluctuations in soil organic matter and nutrient availability. For example, nitrogen limitations could constrain the initial establishment of pioneer vegetation and influence successional dynamics as these plant communities develop. Long-term studies that monitor changes in soil chemistry and vegetation responses could identify successional trends directly correlated to nutrient stock fluctuations.

5. Conclusions

Our study indicates that the intense removal or addition of biomass caused by our treatments did not measurably impact the nutrient pools. These findings suggest that both biomass harvesting and mastication could be utilized as sustainable alternatives for longleaf pine restoration in the region, compared to conventional harvesting methods. Mastication reduced woody debris loading for larger size classes, while simultaneously retaining soil moisture at significantly higher rates compared to other treatments. By exhibiting consistent soil moisture retention rates, even during the driest parts of the growing season, masticated fuelbeds could potentially result in faster vegetation responses and/or a different successional trajectory compared to other treatments. Over time, this will likely result in more rapid nutrient mineralization rates and organic matter incorporation. However, masticated fuelbeds often smolder when prescribed fire is employed, meaning that if there is not a positive vegetation response (e.g., the establishment of fine live fuels), fire implementation could be limited. Without a robust herbaceous understory to carry low-intensity surface fires, masticated sites could experience encroachment by fire-sensitive species. To quantify these changes and understand the influence of silvicultural treatments with regard to restoration, additional management applications and long-term studies need to be applied to comprehend the ecological trends in the region.