These methods are duplicated from the manuscript by Lawrence and Hall in Journal of Environmental Quality. We sampled a traditional corn-soybean (Zea mays-Glycine max) cropping system near Ames, IA, USA (41.982 °N, 93.687 °W). Soils spanned very poorly drained Okoboji series (fine, smectitic, mesic Cumulic Vertic Endoaquolls) in the depression to moderately well drained Clarion series (fine-loamy, mixed, superactive, mesic Typic Hapludolls) in the upland. This depression often floods for days to weeks during wet periods despite subsurface tile drainage and a surface inlet (Martin et al. 2019). Corn and soybean were grown in 2017 and 2018, respectively. Monoammonium phosphate (MAP) fertilizer was broadcast in December 2016 (18 kg N ha-1 and 39 kg P ha-1). Urea ammonium nitrate (UAN) was applied on 4 April 2017 (112 kg N ha-1) and again on 9 June 2017 (67 kg N ha-1). The field was chisel-plowed after corn harvest in 2017 and was cultivated before planting in both years.
Experimental design
Between January 2017 and December 2018, we monitored 8 plots along a linear topographic gradient from the bottom of a depression into the surrounding upland. The transect spanned 120 m with 2.25 m of elevation change. Plots 1–3 were located within the depression and were most likely to pond water, plots 4 and 5 were in the transition zone between the depression and the upland, and plots 6–8 were in upland soils. Lysimeters were installed at 35 cm depth, chosen because soil organic carbon and N, denitrification rate, and extractable nitrate were greatest above 30 cm (McCarty and Bremner 1992; Cambardella et al. 1999). Similarly, most root biomass in Iowa fields was above 30 cm (Nichols et al. 2019), and much of the annual nutrient leaching occurs during spring before roots have established (Daigh et al. 2015). Finally, the 35 cm lysimeter depth was chosen to enable sampling above typical groundwater depth, acknowledging that surface water often ponds from above in depressions due to accumulation of overland flow, and that groundwater rise may sometimes contribute to depressional ponding (Khan and Fenton 1994; Arenas et al. 2018; Schilling et al. 2018). In cases where groundwater rise contributed to lysimeter samples, we would not necessarily expect systematic changes in measured nutrient concentrations, since the groundwater is predominantly derived from nearby infiltration and its chemical composition may be influenced by biogeochemical processes near the lysimeter (Figure S1).
To capture drainage, we used zero-tension lysimeters that diverted infiltration to a buried container. Lysimeters were constructed from 35-cm lengths of 20.3-cm diameter polyvinylchloride (PVC) tubing cut lengthwise to create a half cylinder (Williams et al. 1996; Figure S1). One side was sealed with a halved PVC socket cap and a hose barb outlet was threaded into the base. The top of each lysimeter was installed at 35 cm depth into the sidewall of a soil pit excavated by shovel to ~1 m depth; pits were dug downslope to avoid disturbing overlying soil. Each lysimeter was filled with washed sand and pressed from below into overlying soil (Figure S2). Subsamples of washed sand yielded no measurable nitrate (<0.1 mg N L-1). Quartz wool was installed at the lysimeter outlet, which was connected to a 2 L high-density polyethylene bottle with compression couplings and vinyl tubing (Figure S2). Separate vinyl tubing extended from the bottle to the soil surface, protected by an external PVC tube. Field operations sometimes broke the sampling tubing, and plots 4 and 8 could not be sampled during early summer of 2017, so these data were not reported in 2017. Several additional sampling tubes were destroyed by tillage in November 2017 and these lysimeters were replaced prior to 2018.
Lysimeters were sampled within 24 h of snowmelt or rainfall and at 2-d intervals thereafter using a peristaltic pump until lysimeters did not yield water. A 35 mL subsample was filtered through pre-combusted Whatman GF/F glass fiber filters (0.7 µm pore size) and frozen at -20° C. The P in these samples is interpreted as SRP. Total sample volume was recorded and nutrient yield was calculated as the product of analyte concentration and sample volume divided by lysimeter area (724 cm2). Sample volume was occasionally limited by the collection bottle, which was full for 37 samples. Drainage was likely underestimated in these cases, and we do not report total cumulative nutrient mass yields in this study because of this uncertainty. Another assumption of the method is that soil water generally moves downward; this may be violated in wet conditions when a high water table in the uplands produces groundwater flow into the depressions (Arenas et al. 2018; Schilling et al. 2018). When soils were saturated, our measurements might overestimate true infiltration volume as sampling may have promoted flow into the lysimeter. To provide additional context, we report surface water ponding depth measured during the growing season (mid-May through mid-October) as previously reported by Martin et al. (2019). Briefly, a vented pressure transducer (corrected for barometric pressure) was installed in a stilling well to measure water depth above the soil surface at the depression bottom.
Soil measurements
We collected soil samples to understand how soil nutrient concentrations and redox state related to leaching. Three cores (7.3 cm diameter x 10 cm depth) were collected adjacent to each lysimeter at approximately monthly intervals when soil was not frozen. Soils were collected immediately adjacent to the crop rows, at the midpoint between rows, and at one intermediate location. Values from the three cores were averaged by sampling plot. Six sampling dates from 2017 and eight from 2018 yielded 336 soil samples. We measured Fe(III) and Fe(II) extracted by 0.5 M hydrochloric acid (HCl). The anaerobic metabolic process of Fe reduction is less energetically favorable than denitrification and increased Fe(II) can indicate soils where denitrification was likely to occur or had recently occurred, evidenced by a strong correlation between Fe(II) and denitrification enzyme activity in a previous study (Hall et al. 2016).
Immediately after soil sampling, a subsample was immersed in the field in a centrifuge tube with 0.5 M HCl in a 1:10 soil to solution ratio, solubilizing adsorbed Fe and highly reactive Fe minerals (Thompson et al. 2011) while suppressing Fe(II) oxidation due to low pH. Soil mass was calculated by weighing tubes before and after sampling and correcting for gravimetric moisture measured on separate subsamples. Tubes were processed within 6 h as described below.
We used 2 M potassium chloride (KCl) extractions in a 1:5 soil to solution ratio to quantify ammonium and nitrate. These data were previously published in a study on nitrous oxide emissions (Lawrence et al. 2021) and are used here to interpret nutrient leaching dynamics. The HCl and KCl extraction data were also used as covariates in a published study focused on microbial community analysis (Yu et al. 2021), but spatiotemporal trends were not analyzed. The slurries of soil and HCl or KCl were vortexed for 1 min, extracted for 1 h on a rotary shaker, centrifuged at 10,000 g for 10 min, and solutions were decanted to clean containers for storage at 4° C (HCl) and -20° C (KCl).
Chemical analysis
Nitrate, ammonium, SRP, and Fe(II) and Fe(III) were quantified colorimetrically on a 96-well microplate reader (Biotek Synergy HT, Winooski, VT) with standards in appropriate matrices included on each individual plate (D’Angelo et al. 2001; Doane and Horwáth 2003; Huang and Hall 2017b). SRP could not be measured on six leachate samples with insufficient sample volume.
A subset of lysimeter samples (130 of 180 total) spanning the study period was analyzed for nitrate d15N and d18O at the University of California-Davis using the bacterial denitrification method (Sigman et al. 2001). Isotope ratios are reported using δ notation relative to air for 15N and Vienna Standard Mean Ocean Water for 18O. Bulk δ15N in soils collected at 0–10 cm, 10–20 cm, and 20–30 cm from eight plots spaced equally along the transect was measured using an elemental analyzer and isotope ratio mass spectrometer at Iowa State University. The soil d15N data are discussed in greater detail by Huang et al. (2023).
