Overview and Site Management
We measured fluxes of N2O and
CO2 as well as edaphic and climatic variables
in an agricultural field near Ames, IA (41.98° N, 93.68° W) between
February 2017 and October 2019. Fluxes and were measured at eight
plots along a topographic transect from the center of a topographic
depression into the surrounding uplands. The transect was oriented
perpendicular to the topographic contour and spanned an elevation
change of 2.25 m along 120 m (1). Agricultural management was
typical for the region, with corn (Zea mays)
and soybean (Glycine max) cultivated in annual
rotation with biennial fertilizer application during the corn phase.
Soils in our study system were formed in glacial till and span the
very poorly drained Okoboji series mucky silt loam in the depression
(fine, montmorillonitic, mesic cumulic Haplaquolls) to the
moderately well-drained Clarion series in the adjacent upland
(fine-loamy, mesic typic Hapludolls). The site had sub-surface tile
drainage and a surface inlet at the lowest point in the depression.
In spite of this drainage infrastructure, occasional flooding
occurred in the depression over periods of hours to weeks (2). On 15
December 2016, prior to our first measurements, monoammonium
phosphate fertilizer was spread on the soil surface at a rate of 18
kg N ha-1. Urea ammonium nitrate (UAN)
slurry was injected at a rate of 112 kg N
ha-1 prior to planting corn on 4 April
2017 and 22 April 2019. An additional 67 and 56 kg N
ha-1 of UAN were applied as side-dress
on 9 June 2017 and 30 June 2019. Following corn harvest, soils were
tilled to ~30 cm depth in November 2017 with a combination disc
ripper; no fall tillage occurred after soybean cultivation in 2018
or prior to the final field measurements made in 2019. Soils were
cultivated (~8 cm depth) immediately prior to planting in all three
study years. Chemical pesticides were used for insect and weed
control. The agricultural practices employed at our field site,
including fertilizer type, application rate, and timing, were
typical for the region (3).
Automated Gas Sampling
We utilized dynamic, automated, steady-state chambers to measure
both soil N2O and CO2
flux (1). Each chamber was measured at 4-h intervals (aside from an
initial one-month testing period of 8-h intervals) and a second
chamber was added to each plot in August 2017 and removed in January
2019 (i.e., 16 rather than 8 chambers were used during this period).
Details on analyte gas measurement, calibration, and calculation of
fluxes are described in detail by Lawrence and Hall (1). Briefly,
during each measurement, the chamber lid was closed and ambient air
was pumped through the chamber at a constant rate to achieve
steady-state concentrations of N2O and
CO2 inside the headspace. Chamber inlet and
outlet gases were pumped through gas-tight tubing to an instrument
shed at the edge of the field where the gas analyzers and chamber
control system were located. Fluxes of N2O
and CO2 were calculated as the difference in
mass concentrations between the ambient air entering each chamber
and air measured at the chamber outlet, multiplied by the measured
flow rate. Measurements of N2O and
CO2 were calibrated every 2 h as described in
(1). Chamber lids were opened between measurements (87% of the total
time with 4-h intervals).
Manual Gas Sampling
The automated chambers and tubing were removed during periods of
agricultural management or when flooding exceeded 5 cm depth in the
depression. To fill these measurement gaps, we relied on
supplemental manual chamber sampling at weekly to biweekly intervals
during 2018 and 2019. Manual sampling of trace gas fluxes took place
at 10 plots adjacent to the automated chambers and measurements were
linearly interpolated by elevation to correspond with the eight
automated chamber locations. The manual chambers were fitted with a
vent following the design of Xu et al. (4) to minimize artifacts
from pressure perturbations due to the Venturi effect. Manual
sampling was conducted between 8:00 and 15:00 Central Time with 89%
of sampling conducted before 12:00 to best approximate the 24 h mean
soil CO2 and N2O
emissions (1). The order by which each chamber was sampled was
randomized on each sampling date. Gas samples were collected from
the chamber headspace by piercing a butyl septum with a needle at 0,
5, 10, and 20 min after chamber closure, and were transferred by
syringe for storage in evacuated glass vials for < 7 d prior to
analysis by gas chromatography (Shimadzu 2014A, Waltham MA) using an
electron capture detector for N2O and thermal
conductivity detector for CO2. Gas fluxes
were calculated from the time series of gas concentrations from each
chamber measurement using the extended Hutchinson-Mosier non-linear
model (or linear regression in cases when no valid model could be
fit) as implemented in the HMR package in R (5).
Soil Physical Measurements
Soil moisture was measured adjacent to each automated chamber at
10-min intervals with water content reflectometer probes (30 cm
length; Campbell Scientific 616, Logan, UT) installed from the soil
surface at a 45° angle, thus measuring 0–20 cm depth. Volumetric
water content was derived from the raw reflectometer values using a
soil-specific calibration curve as described in the sensor manual.
Soil temperature at 10 cm depth was measured with temperature probes
(Campbell Scientific 107) at three plots representing the highest,
lowest, and mid-slope locations, respectively. Because no consistent
soil temperature trend was observed across the transect, the average
soil temperature was applied to all plots.
Gap-filling was required for some periods where instruments were not
functional. Soil volumetric water content (VWC) was averaged over
replicate measurements from each plot during the period when two
replicate flux chambers were installed at each plot. When VWC values
from a given plot were missing, we applied the average values of the
two adjacent plots. The remaining missing VWC values where trace gas
fluxes were measured (~400 observations, or 2% of the entire
dataset) were gap-filled using linear interpolation over time across
each individual plot. Average soil temperature values were missing
for ~3,500 observations or 16% of the entire dataset. Soil
temperature was gap-filled using temperature measured at the edge of
the field (~100 m from our plots) at 5 cm depth (6). Missing soil
temperature values were predicted by applying a linear regression
between field and edge-of-field measurements made at 5 cm, which was
the best single predictor of our field measured temperature values
(R2 = 0.92) when both measurements were
available.
Soil Chemical Extractions
Three soil cores (7.3 cm diameter x 10 cm depth) were collected
adjacent to each automated chamber at approximately monthly
intervals during periods when the soil was not frozen. Soils were
collected from locations immediately adjacent to the crop rows, in a
median position between two rows, and one intermediate location. Six
sampling dates from 2017, eight from 2018, and seven from 2019
yielded a total of 504 soil samples. Samples were transported to the
laboratory in a cooler and processed within 6 h.
Soil subsamples were immersed in 2 M potassium chloride
(approximately 1:5 ratio of dry mass equivalent to solution),
vortexed for 1 min to break up aggregates, and extracted for 1 h on
a rotary shaker. Solutions were centrifuged at 10,000 g for 10 min
and the supernatant solution was decanted to a clean container and
kept frozen until analysis. Solution nitrate and ammonium were
analyzed by colorimetric microplate assays (7, 8). To measure net
nitrification and net N mineralization, replicate subsamples from
each field soil sample were incubated in a dark, humidified
environment at lab temperature (~23 °C) for 28 d. After incubation,
these soils were extracted and analyzed using the same methods
described above to quantify net N ammonification as the difference
between pre-incubation and post incubation
NH4
+
concentrations expressed as a daily rate, and net nitrification as
the difference in
NO3
-. To fill gaps
in soil extraction characteristics (all soil N metrics) between soil
collection dates, we relied on linear interpolation. A small number
of gas measurements (~800 or 4% of all gas measurements) were
conducted prior to the first soil collection, N metrics during this
period were back-filled from the first soil collection. No gas
sampling occurred after the final soil sampling date. Two automated
tipping bucket precipitation gauges at the field edge measured
precipitation at 15-min intervals. Soils collected in May 2018 were
further analyzed for pH, texture, bulk soil C, and carbonate
concentration. To measure pH by electrode, soils were vortexed for 1
min using a 1:1 ratio dry soil to water ratio. Soil bulk C
concentration was quantified with a Vario Micro Cube elemental
analyzer (Elementar, Langenselbold Germany). Soil carbonate
concentration was assessed by measuring the
CO2 produced following addition of
hydrochloric acid (9). Soil organic carbon (SOC) was quantified as
the difference between bulk soil C and soil carbonate concentration.
References
1. N. C. Lawrence, S. J. Hall, Capturing temporal heterogeneity in
soil nitrous oxide fluxes with a robust and low-cost automated
chamber apparatus. Atmospheric Meas. Tech. 13,
4065–4078 (2020).2. A. Martin, A. L. Kaleita, M. L. Soupir,
Inundation patterns of farmed pothole depressions with varying
subsurface drainage. Trans. ASABE 62, 1579–1590
(2019).3. P. Cao, C. Lu, Z. Yu, Historical nitrogen fertilizer use
in agricultural ecosystems of the contiguous United States during
1850–2015: application rate, timing, and fertilizer types.
Earth Syst. Sci. Data 10, 969–984 (2018).4. L.
Xu, et al., On maintaining pressure equilibrium
between a soil CO2 flux chamber and the
ambient air. J. Geophys. Res. Atmospheres 111
(2006).5. A. R. Pedersen, HMR: flux estimation with static
chamber data (2020) (October 26, 2020).6. IFIS, Iowa
Flood Information System (2017).7. T. A. Doane, W. R. Horwáth,
Spectrophotometric determination of nitrate with a single reagent.
Anal. Lett. 36, 2713–2722 (2003).8. M. W.
Weatherburn, Phenol-hypochlorite reaction for determination of
ammonia. Anal. Chem. 39, 971–974 (1967).
