Assessment of long-term management impact on soil C dynamics in subtropical grasslands

This research was conducted at the University of Florida Range Cattle Research and Education Center in Ona, Florida (27°23’76”N, 82°56’11”W). The experimental site is characterized by relatively homogenous slope (<5%), and subtropical climate with 10-year average annual precipitation of ~1206 mm and temperature of ~21.5°C. The study was carried out on adjacent fields of three grassland management systems that were consistently maintained for over 20 years. The three management systems constituted a gradient of management intensities ranging from native rangeland (lowest), silvopasture (intermediate), to sown pasture (highest). Prior to the establishment of the adjacent silvopasture and sown pasture, the entire site was native rangeland for livestock grazing. Each collocated ecological unit was replicated twice (6 ha each). The dominant soil series was the same across the sites and consisted of Ona and Immokalee fine sand (sandy siliceous, hyperthermic Typic Alaquods). This soil was developed on parent material of sandy marine deposits (Soil Survey Staff, 1999; NRCS Websoil Survey, 2013).

The native rangeland ecosystem consisted primarily of saw palmetto (Serenoa repens Bartr.) and a wide variety of grass genera including Andropogon, Panicum, Aristida, and Schizachyrium spp. (Kalmbacher et al., 1984). This ecosystem was never fertilized, but it had been subjected to periodic burning (every 3 years), occasional livestock grazing activities (<60 days per year), and herbivory by wildlife, all typical features of rangeland in this region. Each experimental unit was grazed only during winter at rate of 125 animal unit days ha^-1 yr^-1 (a 500-kg animal grazing for 1 day equals 1 animal unit day). Animals were fed a daily supplement of warm-season grass hay and sugarcane molasses at 1.5 to 1.9 kg cow^-1 day^-1 and 0.7 kg cow^-1 day^-1 , respectively, throughout the grazing period.

The silvopasture system was managed for 22 years and consisted of slash pine (Pinus elliottii) trees planted in double-rows (1.2 m along ×2.4 m between rows), and bahiagrass (Paspalum notatum) planted in alleys (12.2 m wide). The vegetation of the silvopasture was established in order of bahiagrass first, followed by slash pine, on a previously native rangeland. Native vegetation was suppressed by burning followed by plowing (~ 45 cm deep) and disking with a dual tandem disk harrow until there was no vegetation on the soil surface. These experimental units received periodic applications of 67 kg N ha^-1 yr^-1 as ammonium nitrate. No fertilizer was applied in the years of 1993-1997, 2000, 2002, 2008, 2009, and 2011. Grazing of the silvopasture began in March 1993, 18 months after planting the trees, and has continued from March-September every year. Each experimental unit was rotationally stocked for 7 months each year, with a 2-week grazing period followed by a 5-week resting period. Stocking rate was 207 animal unit days ha^-1 yr^-1. Animals were supplemented in the pasture with warm-season grass hay and sugarcane molasses at 1.5 to 1.9 kg cow^-1 day^-1 and 0.7 kg cow^-1 day^-1, respectively, from January to April. The management conditions that were applied represent limited inputs of fertilizer and moderate levels of biomass removal through grazing.

The sown pasture system consisted of 32-year-old bahiagrass stand, which was managed to provide forage for grazing livestock. Each experimental unit was stocked at a rate of 360 animal unit days ha^-1 yr^-1. Grazing occurred on a rotational basis for 7 months, with 1-week residence period followed by 1-week rest period between each grazing event. Animal were fed a daily supplement of warm-season grass hay and sugarcane molasses at 1.5 to 1.9 kg cow^-1 day^-1 and 0.7 kg cow^-1 day^-1 , respectively, from January to April. Ammonium nitrate fertilizer was applied at an annual rate of 67 kg N ha^-1 yr^-1 and dolomite was applied in 2001 and 2008 at a rate of 730 and 550 kg ha^-1 respectively. These management conditions are typical of the beef cow-calf production system in Florida.

The study was based on a comparative-mensurative experimental design (Hurlbert, 1984; Arevalo et al., 2009) because the two replicate fields (6 ha each) of each management ecosystems are collocated; in other words, the treatment replicates were not spatially dispersed. This experimental design has an underlying assumption that the soil properties of the three ecosystems (native rangeland, silvopasture, and sown pasture) were similar prior to the conversion and designation of each ecological unit (Hurlbert, 1984; Arevalo et al., 2009). Factors such as the uniformity of initial land-use, the flat terrain, and the limited potential for variation in the sandy-textural class of the soil which were all formed from the same parent material (Kalmbacher et al., 1984; Kalmbacher et al., 1993) lends credence to the assumption that the soils were similar before the management changes were imposed.

Below-ground C sampling and analysis: Soil sampling was conducted in June/July 2012 which corresponds to the period of maximum ecosystem productivity and biochemical cycling rates (decomposition/mineralization). An initial offset (30 m) was established from the edge of each field in order to avoid potential edge effects, and five quadrats (20 m × 20 m), spaced ~75 m apart, were marked out along a diagonal transect within the inner boundary. Four randomly located soil cores (diameter of 2.2 cm) were sampled from each quadrat at soil depths of 0-10, 10-20, and 20-30 cm and composited within a depth for C and N analysis, while one random soil core was sampled in each quadrat for bulk density determination.

Soil samples were air-dried and sieved (2-mm sized sieve), and the modified version of the procedure described by Cambardella and Elliott (1992) was adopted to separate SOC and N into two particle-size fractions: particulate organic C (POC) which corresponded to the > 53µm-sized particles, and mineral-associated C (Cmin) which corresponded to the < 53µm-sized particles. Carbon and N concentrations were determined by dry combustion using a ThermoFlash EA 1112 elemental analyzer. The natural abundance isotope ratios (δ¹³C) of the POC and Cmin fractions were determined on a Thermo-Finnigan MAT Delta^Plus XL Isotope Ratio Mass Spectrometer (IRMS) interfaced via a Conflo-III device to a Costech ECS 4010 elemental analyzer (Costech, Valencia, CA). The results are presented relative to δ¹³C_PDB standard which expresses whether a sample has a higher or lower ¹³C /¹²C isotopic ratio compared to the calcium carbonate standard known as Vienna Pee Dee Belemnite (V-PDB) (Coplen, 1994, Equation 1; Boutton et al., 1998). Carbon derived from initial native rangeland vegetation (C₃-derived C) and from newer sown bahiagrass (C₄-derived C) were calculated based on mass balance equations described by Follett et al. (2009). The soil samples collected at each soil depth interval were dried at 105°C until constant weight and bulk density was computed by dividing the dry weight by the soil core volume. Soil C and N stocks in each ecological unit were calculated based on the C and N concentration and the bulk density at each depth.

Below-ground root biomass samples were collected using the AMS hydraulic powerprobe (Arts Manufacturing and Supply Inc., American Falls, ID) equipped with soil coring drills (diameter of 5 cm). Three soil cores were randomly sampled in each quadrat at the 0-10, 10-20, and 20-30 cm depths. Samples were air-dried and subsequently dispersed in water to separate roots from soil particles through sieving (sieve size = 250µm). Root samples were dried to constant weight at 65°C, and final weight was recorded after 3 days. The dried root samples were combusted at 550°C for 5 hours to determine ash concentration. Carbon and N concentrations were determined by dry combustion using a ThermoFlash EA 1112 elemental analyzer. Root biomass C presented was expressed on ash-free basis.

Above-ground biomass sampling: The above-ground biomass measurement was conducted in mid-summer, in order to estimate the biomass during the period of peak annual primary production. Sampling was strategically conducted to capture the peak production under each management system such that: 1) sown pasture was sampled 5 days after the cows had been moved out of the pastures, i.e. two days before the next herd moved in; 2) silvopasture was sampled ~ 4 weeks after the cattle had been moved out, and 3) native rangeland was sampled after ~3 months of not being grazed by cattle. Different sampling strategies were employed across the ecological units because of the different vegetation composition.

Soil respiration: A modified box exclusion method (Hanson et al., 2000; Luo and Zhou, 2006) was used to partition heterotrophic and autotrophic respiration under each grassland ecosystem. In each sampling location, open-ended fabricated aluminum boxes (30 x 30cm) were installed by carefully digging trenches (with minimal disturbance to the formed soil column) to sever the roots up to 30cm soil depth. The boxes totally excluded roots from growing laterally into the soil column within this soil depth range where over 85% of roots are concentrated. In June 2012, the vegetation within the boxes was clipped to ground level in order to truncate photosynthetic production and avoid allocation of carbohydrate to the roots from aboveground production, thereby leading to eventual death and decay of existing roots. Measurement of respiration was delayed for 6 months to ensure the re-stabilization of soil conditions within the exclusion box, and reduce disturbance effect accruing to the installation.

The weekly in-situ measurements of soil respiration (R_S and R_H) were conducted at two random locations (~10m apart) within each field. Soil CO₂ measurements were conducted weekly (between 8 and 11 am) during the winter (Jan 1 to March 31) and summer (May 15 – August 15) season in 2013. R_S and R_H were measured in-situ with environmental gas monitor (EGM-2) portable infra-red gas analyzer (PP Systems, Amesbury, MA), equipped with a soil respiration chamber (SRC-1). The EGM-2 was recalibrated by the manufacturer to match the operational efficiency and accuracy of EGM-4. Before each weekly measurement event, two (2) polyvinyl chloride (PVC) collars (height = 5cm) were inserted on the soil surface (up to 2.5cm) at each sampling location (one within and one outside the exclusion boxes) to ensure a snug fit of the SRC-1 chamber during measurement, and avoid errors related to potential leakage of respired C. The first PVC (PVC-M) was placed within the exclusion box and designated for measurement of R_H. Given that the vegetation is non-existent and the roots are dead, the CO₂ efflux from this collar is attributable to microbial organisms that are metabolizing the available and accessible soil C within the excluded soil column. The second PVC (PVC-RM) was placed ~0.4m away from each exclusion box to capture the total soil respiration (R_S). The vegetation within the PVC-RM collar was consistently clipped to ground level before each measurement to avoid any influence of shoot respiration. Autotrophic respiration was calculated as the difference between soil respiration and heterotrophic respiration. At each instance of soil respiration measurement, S_Temp (°C) and S_Moist (% volumetric water content) at surface 10cm soil depth were also recorded, using HI 98331 temperature probe (Hanna Instruments, Carrolton, TX) and VG-METER-200 soil moisture meter (Vegetronix Inc, Riverton, UT), respectively.