N fertilizer and harvest impacts on bioenergy crop contributions to SOC

Belowground root biomass is infrequently measured and simply represented in models that predict landscape‐level changes to soil carbon stocks and greenhouse gas balances. Yet, crop‐specific responses to N fertilizer and harvest treatments are known to impact both plant allocation and tissue chemistry, potentially altering decomposition rates and the direction and magnitude of soil C stock changes and greenhouse gas fluxes. We examined switchgrass (Panicum virgatum L.) and corn (Zea mays L.,) yields, belowground root biomass, C, N and soil particulate organic matter‐C (POM‐C) in a 9‐year rainfed study of N fertilizer rate (0, 60, 120 and 180 kg N ha−1) and harvest management near Mead, NE, USA. Switchgrass was harvested with one pass in either August or postfrost, and for no‐till (NT) corn, either 50% or no stover was removed. Switchgrass had greater belowground root biomass C and N (6.39, 0.10 Mg ha−1) throughout the soil profile compared to NT‐corn (1.30, 0.06 Mg ha−1) and a higher belowground root biomass C:N ratio, indicating greater recalcitrant belowground root biomass C input beneath switchgrass. There was little difference between the two crops in soil POM‐C indicating substantially slower decomposition and incorporation into SOC under switchgrass, despite much greater root C. The highest N rate decreased POM‐C under both NT‐corn and switchgrass, indicating faster decomposition rates with added fertilizer. Residue removal reduced corn belowground root biomass C by 37% and N by 48% and subsequently reduced POM‐C by 22% compared to no‐residue removal. Developing productive bioenergy systems that also conserve the soil resource will require balancing fertilization that maximizes aboveground productivity but potentially reduces SOC sequestration by reducing belowground root biomass and increasing root and soil C decomposition.


Introduction
Belowground plant biomass is the primary source of soil organic C and yet is rarely quantified and not dynamically represented in C cycle models. Roots can contribute up to 50% of total plant biomass to belowground C in perennial systems that can extend up to 3 m (Ma et al., 2000;Garten et al., 2010Garten et al., , 2011. Corn (Zea mays, L) can also have a deep root system that contributes 22-50% or more of aboveground biomass to belowground C as roots and root exudates (De Klein et al., 2006;Johnson et al., 2006). Belowground contributions from NT-corn and switchgrass (Panicum virgatum L.) help restore soil quality, reduce erosion potential and increase soil C stocks to marginally productive land (Jin et al., 2015;Stewart et al., 2015b). However, agricultural management practices such as N fertilizer, harvest timing and residue removal can subsequently alter plant allocation and belowground contributions to SOC (Garten et al., 2010).
Annual crops have larger external nutrient requirements compared to perennials, which can remobilize C and N stored in rhizomes and roots (Wayman et al., 2014). Agricultural management including N fertilization and harvest practices can alter plant nutrient allocation between aboveground and belowground plant organs and SOC stocks (Varvel et al., 2008;Heggenstaller et al., 2009;Follett et al., 2012;Ontl et al., 2013). Nitrogen fertilization increases aboveground plant productivity, but many perennial species allocate additional N to aboveground, rather than belowground growth, reducing belowground root production at high N fertilizer rates (Heggenstaller et al., 2009;Garten et al., 2011). Switchgrass harvest removes most aboveground biomass and the timing can significantly alter nutrient translocation belowground (Vogel et al., 2002). Harvesting switchgrass postfrost allows nutrient translocation from the aboveground biomass belowground and decreases N content of aboveground plant material, potentially providing better feedstock quality, but at the expense of the amount of harvested biomass (Vogel et al., 2002;Wayman et al., 2014). N availability can also change SOC stability through the decomposition or priming of existing SOC depending on litter chemistry. N fertilizer and residue quality are two primary controls of litter decomposition. N fertilization increases residue decomposition, but not in all cases (Craine et al., 2007;Hobbie et al., 2012). Under scenarios of N fertilization and higher soil N availability, residues with greater lignin content may be more completely decomposed, providing a lower overall contribution to soil C compared to residues with a lower lignin content (Stewart et al., 2015a). In nutrient-limited situations, plants can increase root exudation, stimulating soil organic matter decomposition (Personeni & Loiseau, 2004;Shahzad et al., 2015) for nutrient acquisition (i.e., Craine et al., 2007). Species with a greater root N content could be expected to decompose and transition more quickly into soil C pools, including the more labile particulate organic matter (POM) fraction, which is comprised primarily of partially decomposed plant material. Crop species, N and residue management all influence the amount and turnover of SOC within these C pools (Camardella & Elliot, 1992).
In this study, we evaluate root-derived contributions to belowground C stocks and decomposition into the particulate organic matter soil C pool under potential cellulosic bioenergy crops no-till (NT) corn and switchgrass. Following conversion from conventionally tilled corn production, the soil profile (0-150 cm) SOC stocks increased under both crops after 9 years compared to baseline (Follett et al., 2012). Here, we evaluate N fertilizer rate and harvest management treatment effects for both crops after 9 years. We hypothesized overall contributions to belowground C pools would be greater under switchgrass, but due to greater N content of root biomass, annual production and death of corn root biomass would provide a greater contribution to particulate organic matter.

Site and experimental design
The long-term switchgrass and maize experiment began in 1998 at the University of Nebraska's Agricultural Research and Development Center (ARDC) (41.151W 96.40N) (details in Varvel et al., 2008;Follett et al., 2012). The climate at the site is mesic with a mean annual temperature of 10.5°C and mean annual precipitation of 765 mm. Field soils were Yutan silty clay loam (fine-silty, mixed, superactive, mesic Mollic Hapludalf) and Tomek silt loam (fine, smectitic, mesic Pachic Argiudoll).
The study was a randomized complete block split-split plot experimental design with two cultivars of switchgrass (Trailblazer and Cave-in-Rock) and NT-corn with three replicates. Due to stand deterioration in Trailblazer, we report data from Cave-in-Rock only.
Three N fertility treatments were randomly assigned within the main plots. Subplots were 30 m long 9 18 m wide and are separated by 15 m wide alleys. From 2000 to 2007, N fertilizer rates were N1 = 0, N2 = 60, N3 = 120 and N4 = 180 kg N ha À1 . Rates on the switchgrass were N1, N2 and N3 and on NT-corn were N2, N3 and N4. The 0N rate for switchgrass was used as a low input treatment only for switchgrass. In 2001, the switchgrass and corn subplots were split lengthwise into 9 m wide sub-subplots for harvest treatments. The two harvest treatments for switchgrass were August or postfrost (PF) harvest date. The NT-corn stover was either removed at 50% (residue removed, or RR) or no residue was removed (NRR).

Aboveground biomass
Aboveground biomass harvesting methods for switchgrass and corn are described in detail in Varvel et al. (2008) and Follett et al. (2012). Briefly, for NT-corn, total aboveground dry matter production (stover + grain) was measured annually by handharvesting a 4.4 m long section of a nonborder, row from each plot after physiological maturity (September or early October). Ears were removed, dried and weighed. Stalks were cut at ground level, chopped, weighed, and a subsample was dried at 60°C until constant mass. After dried ears were shelled, cob mass was added to dried stalk mass to obtain total stover production. Corn grain yields were determined with a plot combine equipped with a weighing unit that harvested the center three rows of each plot. In plots with no stover removal, all stover was left intact following grain harvest. In stover removed plots, stover was removed using a field flail harvester. We report only the 2006 aboveground C yields (grain + stover) to approximate a full accounting with belowground C stocks taken in the spring of 2007.
For switchgrass, the plots were harvested only once a year from a 0.9 to 1.8 m wide swath (varied with harvester used) the full 30 m length of the plots using flail harvesters and associated weighing equipment. At time of harvest, subsamples were collected from each subplot, weighed for moisture content, dried at 50°C for 48 h and reweighed to determine dry matter content. Yields were adjusted to a dry weight basis. The C concentration of the switchgrass samples was determined using near-infrared spectrometer (NIRS) procedures and calibrations (Vogel et al., 2011). Calibration data included samples from 2000 to 2004 from this study reported in Varvel et al.  , 2011). A field flail harvester was used to remove all remaining biomass from the plots immediately following the yield harvest using the same harvest height of 10 cm. We report averaged switchgrass yields from 1998 to 2007 here to integrate lower yields during crop establishment and to integrate extremely low aboveground biomass yields in some plots for 2006 due to drought conditions (Follett et al., 2012). Precipitation was lower than average from 1998 to 2007 (Jin et al., 2015), resulting in low biomass production particularly in 2006 (Follett et al., 2012). Averages include planting year (1998) which reduced average yield.

Soil Sampling
Soils and belowground root biomass were sampled between April and May 2007 before corn planting. Plant material was removed from the soil surface and then, using a flat-bladed shovel, soils were undercut and removed from the 0-5, 5-10 and 10-30 cm depths to preserve soil structure for bulk density analyses. Samples were also collected from the 30-60, 60-90, 90-120 and 120-150 cm depths in May 2007 using a 5.08 cm hydraulic probe (Follett et al., 2012). To minimize plot disturbance, one core was collected for each of the three field replicates between corn rows and switchgrass crowns. Soil bulk densities on all depths were determined using the clod method of the USDA-NRCS National Soils Laboratory methods (USDA-NRCS, 2004). Soils were 2 mm sieved and all >2 mm plant material handpicked from the soil, oven-dried (55°C), subsampled, mechanically ground to pass through a 0.2 mm sieve and stored in glass containers until further analysis. All soils were checked for carbonates and in the very few cases where carbonates existed they were removed with a gentle wash in phosphoric acid prior to analyses for organic C using accepted procedures (Follett et al., 1997). All analyses were on an oven dry weight. The methodology is such that both the isotopic C analyses and the analyses for the total SOC are done at the same time for the same sample.

Root separations, soil fractionation and C analyses
Roots were washed from the bulk soil samples from each depth increment (1 core per field replicate) using a hydropneumatic root elutriator (Smucker et al., 1982). Between two and four replicate subsamples, 50 g of moist soil was washed on 250 lm sieves and the retained root material oven-dried at 55°C. Dried roots were ground to pass through a 0.2 mm sieve for further analysis. These root measurements underestimate corn root biomass due to the spring sampling between corn rows and switchgrass crowns (Allmaras & Nelson, 1971;Johnson et al., 2011), but represent an estimate of standing root stock.
Particulate organic matter (POM) was obtained using a modified method of Cambardella & Elliot (1992). Briefly, after the addition of 0.1 M sodium hexametaphosphate, soil samples were shaken 16 h and separated over a 53 lm sieve (POM > 53 lm). POM was oven-dried at 55°C and ground to pass through a 0.2 mm sieve for C and N analyses.
Total C and N concentrations and d 13 C were determined using a Europa Scientific automated nitrogen carbon analyzer (ANCA-NT) coupled to a Europa 20-20 stable isotope analyzer continuous flow isotope ratio mass spectrometer (Europa Scientific Ltd., Crewe, England). Samples were run in duplicate with an analytical error of 0.14 &.
The d 13 C was calculated using the equation where R sample is the ratio of d 13 C/d 12 C in the soil, and R standard is the ratio of d 13 C/d 12 C international Pee Dee Belemnite (PDB). We partitioned C3-and C4-derived C using the stable isotope mixing model approach (Balesdent & Balabane, 1992):

Statistics
Data were analyzed using a split-split plot design in Proc GLMIX in SAS (Cary, NC . We report least-squared means and standard errors with an n of 3. Differences with a P-value of <0.05 were considered significant, and significance level <0.10 noted in the text. T-tests were used to compare crop effects within N rate for aboveground and belowground C and N.

Aboveground and belowground biomass, C and N
Aboveground biomass in switchgrass ranged from 3.13 to 11.20 Mg ha À1 (Table 1) and 8.30 to 12.43 Mg ha À1 for NT-corn (Table 2). Increasing N fertilizer rate increased aboveground biomass, and biomass C and N of both crops (P < 0.0001 and P = 0.001, Tables 1 and 2). Residue removal had no effect on aboveground C in NT-corn (Table 2). However, postfrost harvest increased switchgrass aboveground biomass C by 14% (Table 1, P = 0.04) and reduced aboveground biomass N by 33% compared to an August harvest (0.033 vs. 0.049 Mg N ha À1 ). There was no significant difference between switchgrass and NT-corn in aboveground biomass C when compared within the same N rate for the 60 (switchgrass = 3.39 Mg C ha À1 vs. NT-corn  3.58 Mg C ha À1 , P = 0.787) and 120 kg N ha À1 rates (switchgrass = 4.55 Mg C ha vs. NT-corn 4.51 kg N ha À1 , P = 0.938). However, NT-corn had 2.4 times more aboveground N stocks at 60 kg N ha À1 and 0.7 times more N stocks at 120 kg N ha À1 compared to switchgrass (P = 0.01, P = 0.03, respectively). Under switchgrass, moderate N fertilizer (60 kg N ha À1 ) maximized belowground root biomass C, 8.66 Mg C ha À1 (0-150 cm, Table 1) but at 120 kg N ha À1 , belowground root biomass C was the lowest of all treatments at 5.43 Mg C ha À1 (Table 1). There was no effect of increasing N fertilizer rate on belowground root biomass C of NT-corn (Table 2). N fertilizer increased belowground root biomass N stocks under both switchgrass (P = 0.04, Table 1) and marginally under NT-corn (P = 0.098, Table 2).
When compared within the same N rate, switchgrass had 6.6 and 4.0 times more belowground root biomass C compared to NT-corn in the 60 and 120 kg N ha À1 rate, respectively (P = 0.001, P = 0.062). Switchgrass had more belowground root biomass N (2.6 times) when compared within N rate to NT-corn in the 60 kg N ha À1 rate (P = 0.007), but not in the 120 kg N ha À1 rate (P = .216). Harvest (postfrost vs. August) had no effect on total belowground root C or N stocks (0-150 cm). Long-term residue removal in NT-corn reduced belowground root biomass C by 37% (P = 0.022) and N by 48% (P = 0.018) compared to noresidue removal (Table 2).

Depth distribution of belowground root biomass C and chemistry
Nitrogen and harvest effects on belowground root biomass C were predominantly observed in the 0-60 cm depths (Fig. 1). Moderate N fertilization (60 kg N ha À1 ) enhanced switchgrass belowground root biomass C in the 10-30 and 30-60 cm depths (Fig. 1a). Increasing N fertilization had little effect on belowground root biomass C under NT-corn, except at 120-150 cm, where belowground root biomass C was greater under fertilizer treatments compared to the 60 kg N ha À1 fertilization rate (P = 0.022, Fig. 1b). Under switchgrass, postfrost harvest increased belowground root biomass C only in 10-30 cm depth (P = 0.030, Fig. 1c). Corn residue removal decreased root biomass C by 30% in the 0-30 cm depth (P = 0.030, Fig. 1d).
Belowground biomass from switchgrass roots had a much greater average C:N ratio (80.2) compared to NT-corn (26.0, P > 0.049), which increased with depth for both crops (Fig. 2a,b). Increasing N fertilizer rate decreased switchgrass root C:N, with the 120 kg N ha À1 rate significantly lower (53.2) than 0 or 60 kg N ha À1 (98.2 and 89.3, respectively) averaged over the soil profile (P < 0.001). The same was observed under NT-corn, with the C:N ratio under 180 kg N ha À1 significantly lower (26.2) than the 60 kg N ha À1 (33.4) averaged over the soil profile (P = 0.006). Postfrost harvest of switchgrass significantly decreased root C:N compared to August harvest in the 30-60 and 60-90 cm depths (Fig. 2c), but not averaged over the soil profile (P = 0.497). Residue removal did not affect C:N under NT-corn (P = 0.467, Fig. 2d). Particulate organic matter carbon The majority of POM-C was observed in the 0-60 cm depths (93-95%) with most treatment differences observed in the 0-5 and 5-10 cm depths (Tables 3 and  4). However, these differences in many cases were large enough to drive soil profile (0-150 cm) changes in POM-C stocks. Surprisingly, there was no significant difference between switchgrass and NT-corn in POM-C (g POM C kg À1 soil) except for the 5-10 cm depth, where switchgrass had 69% more POM-C (P = 0.034, Table 3).
The middle N rate for both crops had the greatest profile (0-150 cm) POM-C (NT-corn, 120 kg N ha À1 5.08 g POM-C kg À1 soil and switchgrass 60 kg N ha À1 6.28 g POM-C kg À1 soil) and POM-C was lowest under the high N rate. These differences were driven by N treatment effects in the 0-5 and 5-10 cm depths for each crop. Switchgrass had the greatest POM-C at the 60 kg N ha À1 rate in both the 0-5 and 5-10 cm depths (3.61 and 1.53 g POM-C kg À1 soil) compared to the 120 N rate (2.08 and 1.21 g POM-C kg À1 soil) (P = 0.032, and 0.011, respectively). Under NT-corn, greater profile POM-C under the middle N rate (120 kg N ha À1 , 5.08 g POM-C kg À1 soil) compared to the 180 kg N ha À1 rate (3.62 g POM-C kg À1 soil, P = 0.035) was also due to smaller POM-C stocks in the 0-5 and 5-10 cm depths under the 180 kg N ha À1 rate.
Residue removal from NT-corn decreased POM-C in the 0-5 cm depth compared to no-residue removal (P = 0.047), resulting in a 22% decrease in POM-C throughout the profile (RR=3.79, NRR=4.89 g POM-C kg À1 soil, P = 0.0376, Table 3). Switchgrass postfrost harvest increased POM-C compared to August harvest in the top three depths although it was only significant in the 10-30 cm depth (P = 0.007), a trend that continued for POM-C throughout the soil profile (P = 0.071, Table 3). Switchgrass had greater POM C:N compared to NT-corn in the top three depths (Table 5).

Belowground root biomass C contributions to SOC
The POM-C stocks under both NT-corn and switchgrass were proportional to belowground root biomass C with R2 = 0.72 for NT-corn and R2 = 0.70 for switchgrass (Fig. 3). However, NT-corn had a three times greater conversion to POM-C per unit root biomass compared to switchgrass (Fig. 3).

Discussion
Perennial bioenergy crops contribute substantially more biomass belowground compared to annual crops, can sequester more soil C and, consequently, have a greater greenhouse gas offset potential compared to conventional corn (Davis et al., 2011). This study illustrates two contrasting scenarios of plant allocation, decomposition and incorporation into soil organic matter. Switchgrass produced more belowground root biomass C (5.5-9.3 Mg C ha À1 ) with a greater C:N ratio compared to NT-corn (1.2-1.9 Mg C ha À1 ) with a lower C:N ratio, reflecting known differences between crops in belowground root biomass production and tissue chemistry (Johnson et al., 2007;Garten et al., 2011;Ontl et al., 2013). This confirmed our hypothesis that overall contributions to belowground C pools would be greater under switchgrass due to greater belowground root biomass. Our measured belowground root biomass underestimated the actual NT-corn root production and reflected a root litter pool due to the spring sampling. Measured root biomass does not account for rhizodeposition, which can contribute an additional 2.5-6 times root biomass C to belowground C stocks (Molina et al., 2001). Despite smaller belowground root biomass C stocks under NT-corn, C was incorporated into the POM fraction three times more per unit of root biomass resulting in no difference in POM-C stocks between the two crops when averaged over the 150 cm soil profile. Faster decomposition of annual corn belowground root biomass into SOC would explain the similar SOC gains under switchgrass and NT-corn after conversion from conventional tillage corn at this site (Follett et al., 2012).

Agricultural management affects switchgrass allocation
Agricultural management significantly altered plant allocation and belowground productivity under switchgrass. The N fertilizer application rate of 120 kg N ha À1 maximized aboveground productivity (4.5 Mg C ha À1 ), but minimized belowground root biomass C (5.4 Mg C ha À1 ). A moderate fertilizer rate of 60 kg N ha À1 maximized belowground root biomass C (8.7 Mg C ha À1 ); a pattern that has been observed in other studies, although at higher N fertilizer rates. Garten et al. (2011) found fertilizer application of 67 kg N ha À1 maximized root biomass measured in April (19 Mg ha À1 ), but was not greater than biomass under the highest N rate of 202 kg N ha À1 . Heggenstaller et al. (2009) found that a N application at 140 kg N ha À1 maximized root production and root N content in Cave-in-Rock in Iowa compared to a N application rate of 220 kg N ha À1 . However, Ma et al. (2000) found that N fertilizer did not impact belowground root productivity at 112 and 224 kg N ha À1 in AL.
Both N fertilizer and harvest timing altered belowground root biomass distribution within the soil profile. Greater belowground root biomass under the 60 kg N ha À1 treatment was due to an increase in the 10-60 cm depth, suggesting increased belowground production for additional nutrient acquisition. Garten et al. (2011) found increasing N demand under higher rates of N fertilization (67 and 202 kg N ha À1 ). Reduced belowground root biomass under the 120 compared to the 60 kg N ha À1 rate suggests allocation of more N to aboveground compared to belowground productivity, a well-documented finding consistent with other studies (Heggenstaller et al., 2009;Garten et al., 2011). Although there was less belowground biomass under the 120 kg N ha À1 rate, the switchgrass root C:N ratio decreased by half compared to the 0N treatment (98.2 vs. 53.2), an effect also observed in other belowground studies (Ma et al., 2000;Garten et al., 2011). Fertilization with 202 kg N ha À1 reduced root C:N ratio (55.1 averaged over live and dead 0-90 cm) compared to 0 kg N ha À1 (107.6) under 5-years-old Alamo switchgrass grown in Tennessee (Garten et al., 2011). The C:N ratio was lower under 224 kg N ha À1 (33) compared to 0 or 112 kg N ha À1 (59) under 3-years-old Alamo switchgrass plants in Alabama (0-3 m, Ma et al., 2000).
Switchgrass harvest after senescence can significantly reduce aboveground nutrient content, providing a better biofuel feedstock (Vogel et al., 2002;Johnson et al. 2014) and allowing nutrient recycling (Zegada-Lizarazu et al., 2012). Postfrost harvest increased aboveground biomass C (3.3 vs. 2.9 Mg C ha À1 ) and decreased aboveground biomass N (43 to 33 kg N ha À1 ). Although we did not observe harvest effects on total belowground root biomass C or N, switchgrass harvested postfrost produced greater belowground root biomass, and 50% more N compared to August harvest in the 10-30 cm depth, suggesting additional C and N storage belowground occurs during senescence. Nutrient storage occurs in crowns (Heggenstaller et al., 2009;Garten et al., 2010Garten et al., , 2011, but our data illustrate that root biomass is also  important for N storage. Cumulative C and N storage from root turnover also contributes to soil C stocks. Garten et al. (2011) found that dead fine roots increased from 5 to 13% of total plant biomass C from April to November indicating substantial mortality over the growing season. Increased root turnover in addition to greater root biomass under the postfrost harvest treatment likely contributed to the observed greater POM-C stocks in the 10-30 cm depth.
Harvest and N treatments that maximized belowground root biomass incorporated more belowground root biomass C into the soil C pool after 9 years. The moderate fertilizer rate of 60 kg N ha À1 maximized POM-C in the top two depths. Postfrost harvest resulted in 18% more POM-C on average compared to August harvest. The POM under switchgrass had a greater C:N ratio in the top 30 cm compared to NT-corn indicating the incorporation of more recalcitrant switchgrassderived C into soil organic matter. The high belowground productivity of switchgrass roots is the primary reason switchgrass can sequester soil C and improve soil quality (Follett et al., 2012;Stewart et al., 2015b). These results show agricultural management practices that maximize aboveground productivity, such as higher N fertilization, may not maximize SOC sequestration through changes in plant allocation.

Root litter decomposition and N effects
Despite switchgrass having 5 times more belowground root biomass compared to NT-corn, there were only slight differences between switchgrass and NT-corn in POM-C confirming our hypothesis that the annual production and death of corn root biomass would provide a greater contribution to particulate organic matter. This difference suggests slower nutrient cycling under switchgrass with the accumulation of perennial belowground root biomass and under corn, faster and more complete decomposition of belowground root biomass. Residue decomposition generally decreases with increasing C:N ratio and switchgrass had more than double the C:N ratio compared to NT-corn. The relatively low C:N ratio of corn belowground root biomass measured here compared to other studies (Johnson et al., 2007) is probably due to decomposition from fall harvest to early spring sampling. Faster decomposition of higher N residues is well documented (Hobbie, 2005;Johnson et al., 2007;Adair et al., 2008;Cornwell et al., 2008). N fertilizer effects on soil C storage are a function of indirect effects on C inputs and decomposition. The highest rate of fertilizer application decreased surface POM-C under both NT-corn and switchgrass. This reduction is most likely due to increased decomposition with additional N fertilizer, because belowground root biomass showed no significant reduction with N fertilizer in the surface depths. However, the highest rate of N addition decreased belowground root biomass C:N, which could further accelerate belowground root biomass decomposition. Initial decomposition rate is frequently positively related to residue N content because microbial substrate use can be N limited at low substrate N concentrations (Hobbie et al., 2012). Subsequent analyses at this site found no difference in NT-corn POM-C stocks across N fertilizer rates (Osborne et al., 2014;Jin et al., 2015). However, differences between our results and theirs likely stem from our values only including C4 POM-C from isotope partitioning, which would identify a smaller C pool from recent crop input.

Residue removal effects in NT-corn
Although residue removal did not significantly affect aboveground productivity, we show reduced belowground root biomass C (37%) and N (48%) relative to no-residue removal which resulted in a profile-wide decrease in POM-C. Continued reductions in belowground root biomass C and N input could lead to reduced C and N stocks. Jin et al. (2015) also found no impact of residue removal on corn yields from 1998 to 2011 on these plots, although earlier samplings did find a small reduction in yield (Varvel et al., 2008). They suggest that soil moisture conservation under the no residue removed treatment during the drought conditions during the early part of the study boosted yields compared to residue removal treatments. Halvorson & Stewart (2015) found a decrease in soil C and no N accrual after 7 years of residue removal in irrigated NT-corn in the semiarid Great Plains despite an increase in grain yield compared to no-residue removal. These results confirm that deterioration in soil quality under residue removal can be evident without crop yield declines (Jin et al., 2015;Stewart et al., 2015b), although grain yield declines under residue removal have been well documented (Wilhelm et al., 2004) .
The reduction in belowground root biomass was primarily observed in the surface depth (0-5 cm) and corresponded to a reduction in aggregate stability, soil microbial biomass and soil quality (Jin et al., 2015;Stewart et al., 2015b). Roots act as nucleation sites for the formation of soil aggregates (Denef et al., 2004;Six et al., 2006) and are the primary source of belowground C in NT systems (Johnson et al., 2006). Many authors document a decrease in soil C pools (aggregates and POM) as a result of stover removal that has typically been assumed to be a result of reduced aboveground C inputs (Blanco-Canqui, 2013;Osborne et al., 2014;Jin et al., 2015). Additional losses of C inputs through reduced root production could further deplete the soil resource.

Belowground C and SOC stocks
Belowground C stocks are important to quantify and this study illustrates two contrasting scenarios that have the potential for substantial belowground contributions to SOC (Follett et al., 2012). Annual NT-corn produced belowground root biomass C proportional to aboveground biomass C with a low C:N ratio and a relatively quick decomposition trajectory into SOC. Under switchgrass, perennial roots built substantial belowground root biomass with a higher C:N ratio and slower decomposition and incorporation as POM-C. Despite these differences in C and N cycling, NT-corn had a greater conversion to POM-C per unit root biomass such that after 9 years, switchgrass POM-C stocks were similar to NT-corn (Table 3). These results may explain why no differences in SOC sequestration between NT-corn and switchgrass were observed in the soil profile after the initial 9 years (Follett et al., 2012). Aboveground biomass production can be a good predictor of SOC sequestration, but previous work on this site found relatively poor relationships with aboveground biomass production (R 2 = 0.39) (Follett et al., 2012). Adding belowground root biomass C increased R 2 of SOC change (0-30 cm, from baseline) to 0.79 for NT-corn and 0.97 for switchgrass (Fig. 4).
N fertilizer can modify nutrient cycling through changing aboveground and belowground biomass allocation patterns and accelerating decomposition directly and indirectly. A moderate fertilizer application (60 kg N ha À1 ) maximized switchgrass belowground root biomass production and POM-C. Higher fertilization rates minimized belowground production and C:N ratio, potentially increasing root decomposition, resulting in the lowest POM-C stocks. Nitrogen fertilization had no effect on the belowground biomass C of NT-corn but the highest N rate (180 kg N ha À1 ) decreased POM-C. Plant allocation patterns and tissue chemistry will have long-term impacts on nutrient cycling in these systems and are rarely investigated over fertility gradients (Garten et al., 2011).
Modeling simulations using the DAYCent model across the Corn Belt region show substantial ecosystem service benefits could accrue from planting perennial bioenergy species on marginal land, including up to a 473% reduction in greenhouse gas emissions and 22% reduction in N leaching compared to conventional corn production (Davis et al., 2011). Despite the potential of perennial bioenergy crops such as switchgrass to contribute to belowground biomass and SOC sequestration, agricultural management such as harvest timing and N fertilization will moderate this effect. Simulation models will need to incorporate feedbacks between plant allocation, belowground biomass chemistry and decomposition to accurately predict land-use effects from bioenergy production.