Effect of contrasting phosphorus levels on nitrous oxide and carbon dioxide emissions from temperate grassland soils

Site description

This experiment was conducted in two long-term P-trial grassland sites (Site A and Site B) situated in proximity (~ 350 m) to each other in the dairy farm at Johnstown Castle, Wexford, Co. Wexford, Ireland (6°49′ W, 52°29′ N). The sites were grazed permanent grasslands before establishment. When the experiment was established in 1995, 16 (10 m × 2 m) plots were formed in each site in a fully randomised block design with four replicates. The two sites established were selected to represent different soil types and drainage classes. Site A is a moderately drained brown earth and site B is an imperfectly drained gley soil³¹. Each year in February, each plot received one of the four phosphorous (P) fertilization rates (16% P superphosphate): 0 (P0), 15 (P15), 30 (P30), and 45 (P45) kg P ha⁻¹ year⁻¹. All plots were initially sown with Lolium perenne and reseeded in 2016 with the same species. However, plant species such as Poa trivialis, Agropyron repens, Trifolium repens were present to a lesser extent. Above-ground biomass is harvested each month between February and August followed by 40 kg N ha⁻¹ fertilizer applications. In the year (2019) of this experiment and the years before, SulCAN as a solid was applied at the first or second week of each month during February-August and potassium (K) as muriate of potash (KCl) was applied in February at a rate of 125 kg K ha⁻¹. SulCAN contains 26.7% N in the form of nitric and ammoniacal nitrogen and 5% water soluble Sulphur. For this study plots receiving P0, P15 and P45 at the two field sites were set up to carry out this experiment. The two sites were selected as they had slightly different soil properties and thus there was an opportunity to consider a soil × treatment effect in the experiment.

Experimental design

Fertilizer N and substrate C were applied on 8 May and 12 June in the experiment undertaken between May and July 2019, which represents the main growing season in Ireland. Within each plot, an area of 1 m × 1 m was selected. Following N fertilizer application (40 kg N ha⁻¹) to all plots, carbon substrate [mixture of glucose (40%), sodium acetate (30%) and methanol (30%)] was applied once within the selected area using a sprayer watering can. Labile C available in animal excreta usually contains carbohydrates, volatile fatty acids, and alcohols³²; as such different carbon substrates were applied to mimic this. Our review of the literature also indicated that C source types could differentially affect denitrifying communities and consequently denitrification rate. Thus, a mixture of three C sources was used to decrease bias of one microbial group over another as a result of single substrate use. Carbon was supplied to alleviate C-limitations of denitrification and nitrification processes as observed by O’Neill et al.²⁹ in soils from this trial and to ensure equal substrate availability across all soil P levels. Equivalent C input rate of 0.63 g C m⁻² day⁻¹ was added to represent a daily rate of plant carbon input from Lolium perenne dominated ecosystem³³. Soil samples were collected on eight occasions throughout the experimental period. Soil was sampled from across each selected area to a depth of 10 cm, sieved through 4 mm sieve and analysed for soil mineral N and microbial biomass.

Soil properties, plant biomass and climate parameters

Physico-chemical soil properties were characterized by taking samples from 10 cm depth from each plot in the two sites before the commencement of the experiment. Soil pH was measured in water (2:1, water volume:soil mass) using Sally pH Auto analyser Dilution System (Gilson 215, Gilson, Dunstable, England). Soil organic matter (SOM) content was determined from mass loss on ignition at 550 °C for 7 h. Total C and total N concentrations were measured using a TrueSpec C/N analyser (TruSpec, LECO Corporation, Michigan, USA). Plant available P, potassium (K), and magnesium (Mg) were estimated using Morgan’s extraction³⁴ and analysed using a Lachat QuickChem 8500 Series 2 Flow injection Analyzer (Lachat, QuickChem, 5600 Loveland, Colorado, USA). Particle size analysis was performed using the Pipette method³⁵, where 2 mm sieved dry soil (20 g) was pre-treated with 6% H₂O₂, 3% NH₄OH, and 5% sodium hexametaphosphate before separating soil aliquots into particle sizes. Water Holding Capacity (WHC) was determined from the mass difference between water-saturated and then overnight dried (105 °C) soil. Bulk density was determined by dividing weight of oven-dried soil by the total soil volume.

To determine the mineral N concentrations, ten gram fresh soil was extracted with 50 mL 2 M KCl (5:1 solution to soil ratio). The supernatant was filtered through Whatman No. 1 filter paper and filtrates were stored in a cold room at 4 °C for about a week until analysis. Ammonium (NH₄⁺) and nitrate (NO₃⁻) concentrations in the extracts were analysed by the Aquakem 600 discrete analyser.

Above-ground plant biomass from each plot of both sites was harvested twice during the experiment period (June 10 and July 11, 2019) to a height of ~ 5 cm using a Haldrup plot harvester. The total harvested biomass weight from each plot was recorded and a 100 g sub-sample was taken for dry matter (DM) analysis. Each fresh herbage sub-sample was weighed and placed in an oven at 70 °C for 3 days, and dry weight of the biomass was determined after re-weighing.

Rainfall records for the experiment period were obtained from a Met Éireann weather observing station located in Teagasc dairy farm in Johnstown Castle, Co. Wexford., situated within a 100 m distance from the experimental sites. Volumetric soil moisture content and temperature was measured to 5 cm depth on individual plots on each gas sampling occasion using a handheld theta probe (WET-2 WET Sensor, Delta-T Devices, Cambridge, England). Water-filled pore space (WFPS) were calculated from the soil moisture values, bulk density of the soils, and soil particle density (2.65 g cm⁻³).

Microbial biomass, glomalin-related soil protein and potential denitrification activity

Soils were analysed for microbial biomass nitrogen (MBN), phosphorus (MBP) and carbon (MBC) using the fumigation extraction method as described respectively in (Brooks et al.^36,37, and Vance et al.³⁸). Five gram fumigated (24 h) and non-fumigated soil samples were extracted with 100 mL 0.5 M NaHCO₃ and analysed for P colorimetrically using an Aquakem 600 discrete analyser (Thermo Electron OY, Vantaa, Finland). In order to avoid the spike readings by the instrument due to the effervescent nature of NaHCO₃, one millilitre of 10% HCl was added to 10 mL extracts and diluted to 50 mL using distilled water. Microbial P was calculated by subtracting the P concentration of non-fumigated samples from fumigated samples, and dividing the result by an extraction factor of 0.4³⁷.

Microbial biomass C and N were determined similarly using chloroform fumigation method with extraction period of 48 h with 0.5 M K₂SO₄³⁸. The extracts of the fumigated and non-fumigated samples were analysed for total C and N using a TOC-L CPH/CPN analyser (Shimadzu, Tokyo, Japan), and the differences, divided by correction factors of 0.45 and 0.54, were used to estimate the microbial biomass C and N, respectively.

Glomalin is a glycoprotein produced by AMF and can be used as an indicator of mycorrhizal colonization in the plant root-soil interface³⁹. Total glomalin-related soil protein (GRPS) was extracted by 90 min of autoclaving (121 °C) of 1 g air-dried soil in 8 mL of 50 mM sodium citrate adjusted to pH 8.0 with HCl⁴⁰. Three additional sequential extractions were performed with the sodium citrate solution by autoclaving for 60 min until no red-brown color was visible in the last supernatant. After autoclaving, the samples were centrifuged at 10,000 revolutions per minute (rpm) for 5 min. The amounts of glomalin in the extracts were quantified using the Bradford dye-binding assay with bovine serum albumin (BSA) as the standard (2 mg mL⁻¹). In a 96-well plate, replicated 200 µL of standard or extracts and 50 µL of dye reagent were added in each well and mixed using a microplate mixer. The Bradford-reactive substance was determined by measuring absorbance at 600 nm using Microplate Reader (Modulus Microplate Multimode Reader, Turner BioSystems, Sunnyvale, California, USA). Sample concentrations were determined using the standard curve. Potential denitrification activity (PDA) was determined using the acetylene inhibition method, modified from Pell et al.⁴¹. Briefly, replicated 20 g fresh soils were added into two identical flasks from a sample of soil. The flasks were then sealed with a rubber stopper and flushed and filled with helium after evacuating the headspace air. In one of the replicas, 10% of the headspaces were removed and replaced by acetylene. All flasks were incubated at 15 °C on an orbital shaker at 175 rpm for 30 min followed by the addition of a nutrient solution containing 75 mmol L⁻¹ KNO₃, 37.5 mmol L⁻¹ Na-succinate, 25 mmol L⁻¹ glucose, and 75 mM Na-acetate. Gas samples were taken from the headspace every 1 h for 5 h. N₂O concentrations were determined using a gas chromatograph (Bruker, Scion 456-GC, Livingston, Scotland), and PDA was calculated from the rate of change of N₂O concentrations over time from acetylene amended flasks.

N₂O and CO₂ flux measurements

Gas samples (N₂O and CO₂ fluxes) were measured before and after the application of N fertilizer and C substrates, with a daily sampling for 10 days directly after C + N additions and 3–4 times a week in the third and fourth week and 2–3 times a week in the subsequent weeks. A rectangular (40 × 40 cm) static collar, made of stainless steel (opaque), was anchored 5 cm deep into the soil within the marked area of 1 m × 1 m in each of the selected plots. During gas sampling, a 10 cm tall chamber lid fitted with two septa on top was placed on the collar lined with neoprene rubber band. To ensure hermetic sealing of the headspace during sampling, the ring area of the collar was half-filled with water, and a 10 kg weight was placed on the top of the lid to compress the seal. Gas samples were collected between 09:30 and 11:30 local time using a 10 mL Luer lock syringe fitted with a hypodermic needle via one of the septa at 0, 20, and 40 min after chamber closure. Prior to transferring the final sample into a pre-evacuated 7 mL glass vial, air in the chamber headspace was mixed by flushing the syringe three times. Gas samples were analysed using a gas chromatograph (Bruker, Scion 456-GC, Livingston, Scotland) fitted with an electron capture detector to analyse for N₂O concentrations and a thermal conductivity detector to analyse for CO₂ concentrations. Daily Fluxes (F) were calculated for each plot using the following equation:

where ∆C is the change in gas concentration in the chamber headspace during chamber enclosure period in ppbv, ∆t is chamber closing period in minutes, so ∆C/∆t is the slope of the gas concentration with time. M is the molar mass of N₂O-N (28 g mol⁻¹) and CO₂-C (12 g mol⁻¹), P and T are the atmospheric pressure (Pa) and temperature (K). Atmospheric pressure values were obtained from the nearby weather station whereas for T, wet sensor values were used. V is the headspace volume of the closed chamber (m³) and A is surface are of the chamber (m³). R is the ideal gas constant (8.314 J K⁻¹ mol⁻¹). Daily flux for each treatment is reported as the average of the replicates.

Cumulative N₂O and CO₂ emissions were calculated over each application period by multiplying the daily N₂O and CO₂ fluxes by the number of days between two consecutive measurements. A summation of the cumulative flux of each application period is reported as the total cumulative flux.

Statistical analysis

ANOVAs with repeated measures were used to test for the C + N addition effect on N₂O and CO₂ emissions, MBC, MBN, MBP, NO₃⁻, and NH₄⁺ with P treatment, site, and day of measurement as fixed effects, and individual plots as a random effect. Two-way ANOVA was applied to test for main and interaction effects of P treatment and site on cumulative N₂O and CO₂ emissions, soil property parameters (Table 1), plant biomass, and GRSP. Prior to analysis, response variables were checked for normality (sphericity for repeated ANOVA) and homogeneity of variance, and log transformed when required. Tukey’s HSD post-hoc tests were conducted to identify pair-wise comparisons of significant effects at P < 0.05. We performed Spearman’s rank correlation to assess the correlations between soil biophysicochemical parameters, plant biomass, and N₂O and CO₂ emissions.

ANOVA analysis was performed using lmer function in lme4 package⁴² within R software. All statistical analyses were conducted using R, version 3.6.0⁴³.

Source: Ecology - nature.com