Maximum soil organic carbon storage in Midwest U.S. cropping systems when crops are optimally nitrogen-fertilized

Abstract

Nitrogen fertilization is critical to optimize short-term crop yield, but its long-term effect on soil organic C (SOC) is uncertain. Here, we clarify the impact of N fertilization on SOC in typical maize-based (Zea mays L.) Midwest U.S. cropping systems by accounting for site-to-site variability in maize yield response to N fertilization. Within continuous maize and maize-soybean [Glycine max (L.) Merr.] systems at four Iowa locations, we evaluated changes in surface SOC over 14 to 16 years across a range of N fertilizer rates empirically determined to be insufficient, optimum, or excessive for maximum maize yield. Soil organic C balances were negative where no N was applied but neutral (maize-soybean) or positive (continuous maize) at the agronomic optimum N rate (AONR). For continuous maize, the rate of SOC storage increased with increasing N rate, reaching a maximum at the AONR and decreasing above the AONR. Greater SOC storage in the optimally fertilized continuous maize system than in the optimally fertilized maize-soybean system was attributed to greater crop residue production and greater SOC storage efficiency in the continuous maize system. Mean annual crop residue production at the AONR was 22% greater in the continuous maize system than in the maize-soybean system and the rate of SOC storage per unit residue C input was 58% greater in the monocrop system. Our results demonstrate that agronomic optimum N fertilization is critical to maintain or increase SOC of Midwest U.S. cropland.

Fig 1. Locations of long-term N fertilization experiments.

Maps show study locations within the most productive region of the U.S. rainfed Maize Belt (eastern Nebraska, southern Minnesota, Iowa, and central and northern Illinois) [20]. a) Mean annual precipitation, b) mean annual minimum temperature, c) mean annual maximum temperature, d) Major Land Resource Areas [21]. All climate data were averaged over 1981–2010 [22]. Cardinal directions for the Major Land Resource Areas are abbreviated in the legend (N, S, E, W, and C for north, south, east, west, and central, respectively).

Fig 2. Cropping system and N fertilizer rate effects on grain yield.

Mean maize and soybean grain yields in response to N fertilizer rate applied to maize in continuous maize and maize-soybean systems. Each of four Iowa study locations is shown on a separate panel (a-d). Grain yields were averaged across 14 (Northwest), 15 (Central and South), or 16 (Southeast) years according to the number of years between soil sampling events. Curves are quadratic-plateau or quadratic models fit to the data (P < 0.01 for all models). Bolded values are the agronomic optimum N rate (AONR) for quadratic-plateau curves. The AONR was set to the highest N rate applied (269 kg ha^-1) for quadratic curves. Error bars represent 95% CIs, calculated using the variability in across-year mean grain yields among replicate plots. Confidence intervals for soybean yields are encompassed by the points.

Fig 3. Cropping system and N fertilizer rate effects on mean annual residue C inputs.

Estimated mean annual inputs of crop residue C in response to N fertilizer rate applied to maize in continuous maize and maize-soybean systems. Each of four Iowa study locations is shown on a separate panel (a-d). Curves are quadratic-plateau or quadratic models fit to the data (P < 0.01 for all models). Error bars represent 95% CIs, calculated using the variability in residue C inputs among replicate plots.

Fig 4. Cropping system and N fertilizer rate effects on soil organic C storage.

Mean (± SE) annual change in surface (0–15 cm) soil organic C (SOC) in response to N fertilizer rate applied to maize in continuous maize (a) and maize-soybean (b) systems. Nitrogen fertilizer rate is expressed as a percentage of the agronomic optimum N rate for each system within each location. Quadratic regression curves are shown for both cropping systems, but the linear and quadratic coefficients were not significant for the maize-soybean rotation (continuous maize: y = -0.15 + 0.0050x – 0.000024x², P < 0.01 for all coefficients; maize-soybean: y = -0.071 + 0.00070x – 0.0000020x², P < 0.05 for intercept, P = 0.28 for linear coefficient, P = 0.66 for quadratic coefficient). For reference, the horizontal dotted lines represent no SOC change.

Fig 5. Cropping system and N fertilizer rate effects on change in the soil C/N ratio.

Mean (± SE) annual change in the surface (0–15 cm) soil C/N ratio in response to N fertilizer rate applied to maize in continuous maize (a) and maize-soybean (b) systems at four Iowa locations. Nitrogen fertilizer rate is expressed as a percentage of the agronomic optimum N rate for each system within each location. Regression lines are shown for both cropping systems, but the slope was not significant for the continuous maize system (continuous maize: y = -0.045–0.00003x, P < 0.01 for intercept, P = 0.71 for linear coefficient; maize-soybean: y = -0.035–0.00012x, P < 0.05 for intercept and linear coefficient). For reference, the horizontal dotted lines represent no change in the C/N ratio.

Fig 6. Relationship between soil organic C storage and residue C inputs.

Mean (± SE) annual change in surface (0–15 cm) soil organic C (SOC) in response to mean (± SE) annual residue C inputs for continuous maize (a) and maize-soybean (b) systems. Solid regression lines apply to the full range of residue C inputs for each cropping system (continuous maize: y = -0.24 + 0.075x, P < 0.01 for both coefficients; maize-soybean: y = -0.20 + 0.048x, P < 0.05 for both coefficients). The dashed regression line for the continuous maize system was fitted using the range of residue C inputs equal to the range observed for the maize-soybean system; the regression lines for the full and truncated data ranges did not differ (P = 0.43). Some horizontal error bars are encompassed by the points. For reference, the horizontal dotted lines represent no SOC change.

Fig 7. Conceptual relationships between N fertilizer input and maize yield, residue production, and residual soil inorganic N.

The agronomic optimum N rate (AONR) is the N rate at which crop yield is maximized. Expected soil organic C (SOC) responses to fertilization of N-deficient maize (below the AONR, grey area) and N-sufficient maize (above the AONR, white area) are shown in bold above the plot.