Experimental design and cover-crop management
The field experiment was set up at Quzhou experimental station (36.87°N, 115.02°E) in Hebei province in November 2019. The site is at an altitude of 40 m above sea level and has a temperate monsoon climate. The annual mean temperature is 13.2°C and the annual mean precipitation was 494 mm (1980 to 2010; range 213–840 mm), with 68% of the annual precipitation falling between June and September (Meng et al., 2012). Precipitation and daily average temperature during the field experiment are shown in Fig. S1 and Fig. S2, respectively. The soil texture of the experiment was sandy loam. The surface 20 cm of the Fluvo-aquic soil had a bulk density of 1.36 g cm− 3, a pH of 7.95, an organic matter content of 16.37 g kg− 1, total N concentration of 0.75 g kg− 1, Olsen-P of 3.02 mg kg− 1and plant-available K (Mc Lean and Watson 1985) of 173 mg kg− 1. These values were determined on one composite soil sample across the whole experimental field before the start of the field experiment.
A single-factor experiment was established to determine the effects of CC types. Four types of CCs were tested: (i) a control with no CC (Fal.); (ii) hairy vetch (Vicia villosa Roth.) (HV); (iii) February orchid (Orychophragmus violaceus.) (OV); (iv) a mixture of hairy vetch (Vicia villosa Roth.) and February orchid (Orychophragmus violaceus.) (HO). The experiment was a completely randomized design with four replicates in blocks. The surface area of the elementary plot, containing 20 rows for each treatment, was 24 m2 (6 m×4 m). To avoid plant–plant competition effects between adjacent treatments, only the twelve rows in the middle of the plot were harvested and used for soil measurements.
Species grown as sole crops were sown at densities recommended by cover crop seed companies, breeders, and agricultural advisors. The seeding rate of hairy vetch and February orchid was 60 kg ha− 1 and 30 kg ha− 1, respectively. In mixture, sowing densities were half of the corresponding sole crop density of each species (50% hairy vetch: 50% February orchid). Seeds of both species were mixed before sowing to ensure that they were mixed in the row. Sowing was carried out with a row width of 20 cm and a sowing depth ranging from 1.5 to 2 cm on 22 September 2019. The proportion of each sown species was controlled during the cover crop emergence phase. No fertilizer, irrigation or herbicides were applied to CCs throughout their growing period. On bare soils, only manual weeding was performed.
Preparation and isotope labeling of fertilizers
The microplot experiment was set up within the respective treatments of the large-plot experiment (6×4 m per plot replicated in three blocks) on 21 October 2019. Galvanized metal frames confining an area of 1.0 ×1.2 m (1.2 m2) were driven 20 cm into the soil leaving 3 cm above the ground. Six rows of catch crops were covered in each microplot. Each plot had one microplot, and hence there were four replicates for each type of CCs.
The 15N-labeled urea (3.66 atom%15N) was mixed with 1 kg of soil from each 15N microplot and uniformly spread by hand; then the 0 ~ 20 cm soil layer was tilled using a shovel to achieve a uniform application of 15N fertilizer. A 15N-labeled urea (3.66 atom%15N) solution was applied as a tracer to all microplots on the soil surface along the crop rows using a pipette. The total application of 10 kg N ha− 1 was split into two doses applied on 20 March and 27 March 2020, respectively. This was done to maximize crop uptake and minimize leaching of 15N. Weeds emerging in and around the microplot area were killed on several occasions but left on the soil surface. The CC tops in all microplots were harvested by hand-cutting at the soil surface on 4 May 2020 and then returned to the field area of metal frames on 15 May 2020.
Crop planting and fertilization
In the large-plot experiment, all CCs tops were manually returned to the field on 15 May 2020. Then all treatments were planted with spring maize. Before planting, 6-leaf (V6) stage to silking (R1), and R1 to physiological maturity (R6), optimal N rate was calculated by subtracting the measured soil nitrate content in the root zone (0–60 cm from V6 to R1 and 0–90 cm from R1 to R6) from the corresponding N target values (185 and 160 kg ha− 1 for the two growing periods, respectively). When calculating optimal N rates, to ensure that each plot obtained sufficient N, replicated plots with too high NO3–N were omitted. If the NO3–N in all replication plots exceeded the target N, 30 kg ha− 1 N fertilizer was applied to ensure normal maize growth. Nitrogen fertilizer for all treatments was applied in the form of urea in split doses, as described in Table 1. Together with the urea, 20 kg ha− 1 P as superphosphate and 75 kg ha− 1 K as potassium sulfate were broadcasted and incorporated into the upper 0–30 cm soil layer by rotary tillage before planting. At the V6 and R1 stages, urea was top-dressed by furrow application. The modern stay-green maize hybrid used was Denghai 605. Spring maize was sown on 25 May 2020, with a row spacing of 60 cm and plant spacing of 22.5 cm, ∼75,000 plants ha− 1. The grains were harvested at the R6 stage when the black layer occurred; the harvest date was November 23 in 2020 for all treatment. Pesticides were applied in case there was a need for managing disease, weeds, and insects during the two growing seasons.
Table 1
Nitrogen (N)-fertilizer application rate (kg ha− 1) in four treatments. Fal., fallow-spring maize cropping system; HV, hairy vetch-spring maize cropping system; OV, February orchid-spring maize cropping system; HO, hairy vetch/ February orchid mixture-spring maize cropping system. V6, 6-leaf stage; R1, silking stage.
Treatment | Total | Before planting | V6 | R1 |
Fal. | 273 | 30 | 175 | 67 |
HV | 191 | 30 | 161 | 0 |
OV | 248 | 45 | 203 | 0 |
HO | 197 | 30 | 167 | 0 |
In the microplot experiment, the planting density and field management practice was consistent with the large-plot experiment.
Sampling and analyses
At the end of the growing season (5 May), the aboveground biomass of the CCs was sampled outside the microplots to determine dry matter yield. Here, three 80 cm × 100 cm quadrats were collected from each plot. In the microplots, all the CC tops were harvested and oven-dried at 60°C (48 h) to determine dry matter content and then cut into small pieces (ca. 0.5 cm). The pieces were randomly sampled from each microplot to determine the 15N content of CC biomass.
The total N and atom fraction 15N of plant samples were analyzed at UC Davis Stable Isotope Facility, using a PDZ Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20–20 isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK).
At the R6 stage of spring maize, plants in an area of 5.4 m2 (3 × 1.8 m [three rows]) in the middle of each plot were harvested and threshed, and the grains were dried to determine the grain dry matter. At the V6, R2, and R6 stages, three representative adjacent plants from each plot were cut at the soil surface and separated into leaves, stems (leaf sheaths, tassels, husks and either cobs at R6 or ear shoots at R1), and grains (at R6) to determine the aboveground biomass. All plant samples were dried at 70°C in a forced-draft oven until they reached a constant weight and then weighed to calculate the biomass.
Indicators used to characterize N recovery from fertilizers in the soil-plant system
The recovery rate of the applied 15N enriched fertilizer, RF (%), in the above- or belowground biomass of a CC was calculated as follows:
RF (%) = [(\(\frac{\text{EA of CC}}{\text{EA of fertilize}\text{r}}\)) × CC-N / fertilizer N] × 100, where EA is the excess atom fraction 15N of a sample; CC-N is the total N uptake in aboveground biomass or roots (kg N ha− 1) and the fertilizer N was 10 kg N ha− 1.
The N fixation by CCs was calculated by referring the mean EA of the FR and GR. Therefore, the percentage of N in CC derived from the atmosphere, Ndfa (%), was calculated using the following equation (Huss-Danell and Chaia, 2005):
Ndfa (%) = [1 - (\(\frac{\text{EA of CC}}{\text{mean }\text{EA of non-CC}}\))] × 100, and the amount of biologically fixed N allocated in CC aboveground biomass or roots, N fixation (kg N ha− 1), was:
N fixation = Ndfa (%) / 100 × CC-N
The amount of plant N derived from the soil N pool, Nds (kg N ha− 1) represents the ability of a CC to uptake N from the soil. The Nds included N taken up from both the applied 15N fertilizer and the native soil N pool. Thus, Nds of the non-CC was identical to the total N uptake, while for CCs it was calculated as:
Nds = [1 – Ndfa (%) / 100] × CC-N
Calculation of N input, N output, and N balance
For N input, six indicators were considered: N from fertilizer and irrigation, atmospheric deposition, non-biological fixation, biological fixation, and seeds.
N fertilizer (urea- 46% N) application is shown in Table 1. The amount of irrigation water (1.35 mg L− 1) was 269.6 mm in all cropping systems.
Atmospheric N deposition (dry and wet): This refers to the process of gaseous and particulate N components transported from air to the surfaces of aquatic and terrestrial landscapes (Anderson and Downing, 2006). The samples of atmospheric N deposition were collected by an APS-II sampler, and then measured in the lab. During the whole growth period, the atmospheric N deposition was 60.9 kg N ha− 1.
Non-biological fixation: For nonlegume crops, the value ranges from 4.5 to 20 kg ha− 1 yr− 1 (Bouwman et al., 2005). In this study, we used a value of 15 kg ha− 1 yr− 1 for non-biological fixation (Liu et al., 2008).
Seeds: The maize and CCs samples were collected and weighed. The N content was analyzed by an elemental analyzer (Flash 2000; Thermo, Waltham, MA, USA). N input from seeds was calculated according to the sowing amount (kg ha− 1) and N content (g kg− 1).
To calculate N output, crop uptake, ammonia volatilization (NH3), and N2O, NO, N2 emissions, and NO3− leaching were considered.
Crop N uptake: Mazie was harvested, and the grain was separated from the straw. Crop N uptake by maize was calculated according to the grain/straw yield (kg ha− 1) and N content (g kg− 1).
Ammonia volatilization (NH3), N2O loss, NO loss, N2 loss and NO3− leaching loss: This study selected the latest papers that are similar to the production environment of Quzhou Experimental Base to estimate the loss of active N in the field. N2 and NO emissions are estimated to be equal and seven times the direct N2O emissions (Yan et al., 2013). The calculation is presented below.
NH3 = 6.25%×Uinput (Xie et al. 2015)
N2O = 0.31%×Uinput (Shi et al. 2013)
NO3− leaching = 27.22%×Uinput (Ning et al. 2012)
where Uinput is the amount of urea (kg N ha− 1).
Carbon footprint
To evaluate the carbon footprint (CF) of different winter CC-spring maize cropping systems, a cradle-to-farm gate life cycle analysis (LCA) approach was adopted. The system boundary of GHG emissions considered the periods from production and transport of agricultural inputs and field production processes to the farm gates (Fig. 1). The boundary of GHG emissions include 1) production, transport, and use of agricultural resources (fertilizers, seeds, pesticides, diesel); 2) labor input for farm operations; 3) electricity for irrigation; 4) N2O emissions, and volatilization of NH3 and leaching of NO3− in the field during crop growth (Ntinas et al. 2017). In our study, the CF per unit of area (CFGHG), the CF per unit of economic benefits (CFEB) and the CF per unit of yield (CFY) were calculated to assess the CF of the four cropping systems. The formulas in this study are as follows (Yang et al. 2014):
CF GHG = \(\sum _{i}^{n}\)(AIi ×EFi) + N2Ofield
CF Y =\(\frac{CF\text{GHG}}{{\sum }_{i}^{n}Grain Yield\text{j}}\)
CF EB =\(\frac{CF\text{GHG}}{EB}\)
where CFGHG are the CF per unit of area (kg CO2-eq kg− 1) and CFY and CFEB are the CFGHG per unit grain yield per year (kg CO2-eq kg− 1) and per unit EB per year (kg CO2-eq USD− 1), respectively. 1 United States dollar (USD) is equal to 6.67 Chinese Yuan in this experiment. Grain yieldj represents the amount of grain yield (kg ha− 1) of spring maize. AIi is the amount of each agricultural input applied to one cropping system, and EFi is the GHG emissions factor for each agricultural input. The detailed amounts of each agricultural input in the four cropping systems are presented in Table S1, and the GHG emissions factors for all agricultural input are shown in Table S2.
Statistical analyses
After verifying the homogeneity of error variances, analysis of variance (ANOVA) was performed using SAS software (ver. 6.12; SAS Institute, Cary, NC, USA). A one-way ANOVA model was used to assess the overall variability in grain yield, partial factor productivity of N fertilizer (PFPN), the amount (Nplant−residue) and the proportion of N derived from residues, N balance. Differences were compared using the least-significant difference (LSD) test at the 0.05 probability level in SAS.