Global Change Biology Global Change Biology (2014), doi: 10.1111/gcb.12738

Intercropping enhances soil carbon and nitrogen WEN-FENG CONG1,2, ELLIS HOFFLAND3, LONG LI1, JOHAN SIX4, JIAN-HAO SUN5, X I N G - G U O B A O 5 , F U - S U O Z H A N G 1 and W O P K E V A N D E R W E R F 2 1 College of Resources and Environmental Sciences, China Agricultural University, Beijing 100193, China, 2Centre for Crop Systems Analysis, Wageningen University, Wageningen, 6700 AK, The Netherlands, 3Department of Soil Quality, Wageningen University, Wageningen, 6700 AA, The Netherlands, 4Institute for Agricultural Sciences, Department of Environmental Systems Science, Swiss Federal Institute of Technology, ETH-Zurich, Zurich, CH 8092, Switzerland, 5Institute of Soils and Fertilizers, Gansu Academy of Agricultural Sciences, Lanzhou 730070, China

Abstract Intercropping, the simultaneous cultivation of multiple crop species in a single field, increases aboveground productivity due to species complementarity. We hypothesized that intercrops may have greater belowground productivity than sole crops, and sequester more soil carbon over time due to greater input of root litter. Here, we demonstrate a divergence in soil organic carbon (C) and nitrogen (N) content over 7 years in a field experiment that compared rotational strip intercrop systems and ordinary crop rotations. Soil organic C content in the top 20 cm was 4%
1% greater in intercrops than in sole crops, indicating a difference in C sequestration rate between intercrop and sole crop systems of 184
86 kg C ha1 yr1. Soil organic N content in the top 20 cm was 11%
1% greater in intercrops than in sole crops, indicating a difference in N sequestration rate between intercrop and sole crop systems of 45
10 kg N ha1 yr1. Total root biomass in intercrops was on average 23% greater than the average root biomass in sole crops, providing a possible mechanism for the observed divergence in soil C sequestration between sole crop and intercrop systems. A lowering of the soil d15N signature suggested that increased biological N fixation and/or reduced gaseous N losses contributed to the increases in soil N in intercrop rotations with faba bean. Increases in soil N in wheat/maize intercrop pointed to contributions from a broader suite of mechanisms for N retention, e.g., complementary N uptake strategies of the intercropped plant species. Our results indicate that soil C sequestration potential of strip intercropping is similar in magnitude to that of currently recommended management practises to conserve organic matter in soil. Intercropping can contribute to multiple agroecosystem services by increased yield, better soil quality and soil C sequestration. Keywords: ecosystem services, functional complementarity, intercropping, plant diversity, plant productivity, root biomass, soil carbon, soil nitrogen Received 29 March 2014 and accepted 4 September 2014

Introduction Ecosystem functioning and ecosystem services are inextricably linked to biodiversity (Cardinale et al., 2012). Biodiversity affects the efficiency of primary production, resource capture, decomposition and recycling of biologically essential nutrients. Loss of biodiversity, conversely, reduces this efficiency. Functional complementarity between plant species in grassland ecosystems promotes productivity as well as stocks of carbon (C) and nitrogen (N) in soil (Fornara & Tilman, 2008; Steinbeiss et al., 2008). The positive effect of species diversity on soil C and N stocks in natural grasslands has been attributed to enhanced belowground input of organic matter derived from greater root production in Correspondence: Fu-Suo Zhang, tel. +86 10 62732499, fax +86 10 62731016, e-mail: [email protected]; Wopke van der Werf, tel. +31 317 485315, fax +31 317 485572, e-mail: wopke.vanderwerf @wur.nl

© 2014 John Wiley & Sons Ltd

more diverse plant communities (Fornara & Tilman, 2008; Steinbeiss et al., 2008; Cong et al., 2014). The greater belowground biomass production is associated with the presence of complementary plant functional groups, such as C3 forbs, C4 grasses and N-fixing legumes. Intercropping is the simultaneous cultivation of two or more crop species in the same field (Vandermeer, 1989). Intercropping introduces greater plant diversity and functional complementarity in arable crop systems. Intercrops usually give higher yields per unit area than sole crops as measured by the land equivalent ratio, the relative area of land that would be required to produce the same yields as a unit area of sole crop (Willey, 1979; Vandermeer, 1989; Lithourgidis et al., 2011). The higher productivity in intercropping has been attributed to complementarity between crop species in temporal and spatial patterns of resource acquisition (Hinsinger et al., 2011), facilitation (Zhang & Li, 2003; Hauggaard-Nielsen & Jensen, 2005), and/or reduction in the impacts of 1

2 W - F . C O N G et al. pests, diseases and weeds (Liebman & Dyck, 1993; Trenbath, 1993; Zhu et al., 2000). Intercropping is a potential strategy for ecological intensification of agriculture as it can strengthen regulating and supporting ecosystem services (Bommarco et al., 2013). Despite a large intercropping literature, little information is available on the long-term effects of intercropping on the soil. Soil C and N levels are key quality parameters for agricultural soils because they are positively linked to the supply of nutrients to the crop, water-holding capacity, workability, and resistance to soil compaction, erosion and surface crusting (Weil & Magdoff, 2004). Enhanced C sequestration in soil is of interest to mitigate anthropogenic atmospheric CO2 increase (Lal, 2004). Because of greater productivity per unit land, we expect that intercropping can enhance litter input into the soil, and hence promote the build-up of organic matter and sequestration of C. Indeed, recent studies suggest greater input of C into the soil through root residues in intercrop systems as compared to sole crops (Ghosh et al., 2006; Yang et al., 2010; Li et al., 2011a). However, litter diversity enhances decomposition in some systems (Cardinale et al., 2011) which could prevent accumulation of C in the soil despite greater input; hence a hypothesized effect of intercropping on soil C would need empirical testing. Increases in soil C by intercropping have to date not been demonstrated. Agroecosystems differ in many respects from natural systems, for which the effects of diversity on productivity and sequestration of C and N have been well documented (Hooper et al., 2005; Tilman et al., 2006). First, they consist mainly of rotated annuals whereas the studied natural ecosystems are mainly perennial systems such as grasslands (Cardinale et al., 2012). The role of soil pathogens in annual crop systems is mitigated by rotation (Peters et al., 2003) and roots of annuals decompose faster than those of perennials (Wardle et al., 1997). Furthermore, aboveground residues are partly or wholly removed in agriculture, reducing the input of litter into the soil. Finally, high N inputs in agroecosystems can suppress biological N fixation by legumes (Salvagiotti et al., 2008), thereby potentially reducing complementarity effects. It is therefore impossible to extrapolate the findings in natural grasslands to agroecosystems. Candidate mechanisms for enhanced N storage through intercropping are, first, enhanced biological N fixation by legumes when intercropped with cereals (Li et al., 2009) and, second, improved N capture in mixed crops as a result of complementarity in foraging strategies in space (soil profile) and time (growth period of the crop) (Vandermeer, 1989; Li et al., 2005; Lithourgidis et al., 2011). On the other hand, intercrops may have greater crop N removal resulting from higher yields as compared to sole crop systems. There are no reports on

long-term effects of intercropping on N storage in soil. Some work in this area has been done (Myaka et al., 2006; Snapp et al., 2010), but the duration of those studies was too short to convincingly demonstrate sequestration of organic C and N. In the studies reported here, we determined soil C and N content after 7 years in a field experiment that compared three strip intercrops (maize/wheat, maize/ faba bean and wheat/faba bean) and three sole crops (maize, wheat and faba bean). Both the intercrops and sole crops were grown in two-year rotations, in accordance with farmer practise. In two ancillary experiments, we determined root biomass, as a proxy for belowground productivity, in strip intercrops of maize with wheat or faba bean and compared it to root mass in the sole crops. We tested three hypotheses: (i) Root mass is greater in intercrops than in sole crops; (ii) Intercropping results in soil C sequestration over time; and (iii) Intercropping results in soil N sequestration over time.

Materials and methods The formulated hypotheses were tested in three field experiments. First, C and N content in soil was determined after 7 years of strip intercropping or sole cropping in a long-term experiment that was initiated in 2003 and is still ongoing (Li et al., 2007). Secondly, root mass of wheat, maize and faba bean in sole crops and strip intercrops were determined in two single-year field experiments in 2008 and 2011. All experiments were conducted at Baiyun Experimental Station (38°370 N, 102°400 E) in Gansu province, Northwest China.

Long-term experiment on soil C and N content A long-term field experiment (LTE) was established in 2003 (Li et al., 2007). Average annual temperature at the site is 8.9 °C. Climate is arid with a total yearly rainfall of 168
8 mm and potential evaporation of 2021 mm. Soil texture is a sandy loam from 0 to 71 cm soil depth, clay loam from 71 to 106 cm, silty clay loam from 106 to 144 cm, and sand below. Soil pH was 8.2, and Olsen-P 21.2 mg kg1 in 2003. Three crop species (maize, Zea mays; wheat, Triticum aestivum; and faba bean, Vicia faba) were grown as sole crops and as two species strip intercrops. Both the intercrop and sole crop systems were grown as two-year rotations, to avoid problems associated with continuous cultivation of a crop species and in accordance with farmer practises in China (Fig. 1a). Sole crop rotations were maize/wheat, maize/faba bean and wheat/faba bean. Intercrop rotations had the same three species combinations. Each crop species was grown in 80 cm wide strips, and the position of the two species were alternated between years (Fig. 1b). For instance, in the wheatmaize intercrop, maize was grown in strips in which wheat had been grown the year before, and wheat was grown in strips in which maize was the previous crop. Therefore, both

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

INTERCROPPING ENHANCES SOIL C AND N 3 (a)

(b)

Fig. 1 Overview of the six crop systems in the long-term experiment (2003–2010) on soil C and N content in strip intercrops and sole crops (a) and placement of crop strips in intercrop plots in alternate years (b). Panel (b) illustrates that the intercropped species swap position from one year to the next. Such a ‘small rotation’ is a common practise in China, and provides the benefits of crop rotation even in intercrops. Sole crop systems were also rotated.

the sole crop and intercrop systems were two-year rotations in the temporal sense. The key difference between intercrop and sole crop systems is the species interaction in space. In sole crop systems, there is no such interaction, whereas in strip intercrops, there is intimate interaction between crop plants growing in the border rows of the strips, to which the plants in the border rows show responses in above- and belowground architectural traits (Li et al., 2001; Zhang & Li, 2003; Zhang et al., 2007; Zhu et al., 2014). Wheat and faba bean were sown in late March of each year (Fig. 2). Wheat was harvested early July and faba bean late July. Maize was sown in mid-April and harvested early October, 3 months later than the other two species. Intercrops with maize are therefore relay systems with a considerable difference in growing period between the two species, whereas the wheat-faba bean system is characterized by almost complete synchronicity between the two species. Row distance was 40 cm in maize, 13.3 cm in wheat and 20 cm in faba bean, both in sole crops and intercrops. Plant distance in rows was 25 cm in maize and 20 cm in faba bean. The seed rate of wheat was 675 seeds m2. A maize strip in an intercrop consisted of two rows of maize at 40 cm row distance, a wheat strip consisted of six rows of wheat at 13.3 cm row distance, and a faba bean strip consisted of four rows at 20 cm row distance (Fig. 1b). The distance between the border rows of strips of different species was calculated by summing half the interrow distances within the strips for each species. For instance, in the case of maize/ wheat, the distance between adjacent maize and wheat rows

Fig. 2 Seasonal schedule of crop growth, fertilization and irrigation in the long-term experiment, and two root sampling occasions in the single-year experiments. Shaded areas represent growing periods (sowing to harvest) of crop species. Large arrows (F) indicate times of fertilizer application, and small arrows (I) indicate timing of irrigation. The root sampling occasions are indicated with crosses. Modified after Li et al. (2011b, Fig. 1)

was 40/2 + 13.3/2 = 36.65 cm. Accordingly, all intercrops followed a replacement design where the sum of the relative densities of the two species (density in intercrop divided by density in sole crop) equals 1 (e.g., Zhang et al., 2007). In the case of sole crop rotations, only one phase of the rotation was present in any year (Fig. 1a). In the case of intercrops, there is no ‘phase of the rotation’ because both component species in a system are present in each year. Phase of the rotation is considered of minor influence on soil properties over a sequence of years, but the experimental design does not allow analysing this premise. Plot size was 5.6 m row length by 8 m width,

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

4 W - F . C O N G et al. covering 5 strips of both species in the intercrop plots. The field experiment was laid out as a completely randomized block design with three replicates. All management such as tillage, fertilizer and irrigation was applied uniformly across all treatments to avoid biasing the results. Each plot received 225 kg ha1 root biomass N as urea and 40 kg ha1 P as triple superphosphate annually with twothirds of the amount applied at moldboard tillage (down to 20 cm depth) just before sowing of wheat and faba bean in late March. The remainder was top-dressed at wheat flowering in late May. Irrigation was given seven times per year, five times during the growth periods of all three crops, and two times after harvest of wheat and faba bean (Fig. 2). Aboveground crop residues were removed after harvest for use as feed or fuel. Prior to sowing in 2003, a bulk soil sample was collected by taking cores 0–20 cm deep across the field. On 1 July 2010, soil samples were taken in each plot, when wheat was at the dough stage (White & Edwards, 2008), maize was in stage V10 (Abendroth et al., 2011), and faba bean was in pod-filling stage (Knott, 1990). Cores were taken using an auger (35 mm diameter) down to 1 m depth and subdivided in 0–20, 20–40, 40–60, 60–80 and 80–100 cm. Three cores per plot were taken in sole crop plots and bulked per depth. In intercrop plots, three subsamples of three cores each were taken in each plot, two from the crop strips, and a third one from the transition zone. Samples in the crop strip were taken in the middle between the maize rows in maize, in the middle between the 2nd and 3rd row in the faba bean strip, and in the middle between the 3rd and 4th row in the wheat strip. Soil Organic Carbon (SOC) and Soil Organic Nitrogen (SON) contents at the three locations within each intercrop plot were averaged after initial analysis demonstrated no significant differences between the three locations per plot. Soil samples were transported to the lab, air-dried for 3 days, and sieved over a 1 mm mesh sieve to remove plant debris. Samples were treated with 0.5 M HCl (10 ml g1 soil) to remove inorganic C (Midwood & Boutton, 1998). After drying, the acid-treated soil samples were ground in a ball mill. Acid-washing removes mineral N from the soil sample. Therefore, the analysis method results in quantification of organic N in the soil. SOC and SON were determined in duplicate using C/N analyzers in two chemical laboratories: C/N analyzer (Vario Macro, Elementar, Hanau, Germany) at the Stable Isotope Lab of China Agricultural University, and C/N isotope ratio spectrometer (PDZ Europa Integra, Cheshire, United Kingdom) at the Stable Isotope Facility of the University of California. Soil d15N was determined in the latter. The results on SOC and SON contents were pooled for data analysis, because systematic differences between the two determinations were not found and pooling allowed for greater accuracy. Part of the results on SOC (0–20 cm depth) were further validated against results from wet chemical oxidation (Kurmies) at Wageningen University which confirmed the results from the C/N analysers (cf. Walinga et al., 1992). To measure soil bulk density, six pits were dug, two in each of the three blocks of the experiment, and one in each crop system. Soil samples were collected by gently pushing 5 cm

diameter cylinders (length: 5.1 cm; volume: 100 cm3) laterally into the side of the pit at depths of 10 and 30 cm. Soil samples were oven-dried (105 °C) for 3 days before weighing. Soil organic C, Soil organic N, C/N ratio and d15N were statistically analysed using four-way repeated-measures ANOVA in SPSS 20.0 (IBM Corporation, Armonk, NY, USA) with depth as within-subject factor, and block, crop system (sole crops vs. intercrops) and crop combination (maize/wheat, maize/faba bean or wheat/faba bean) as between-subjects factors. SOC, SON and C/N ratio were further analysed separately for each soil layer using three-way ANOVA, with block as a random factor, and crop system and crop combination as fixed factors.

Single-year experiments on above- and belowground biomass Aboveground biomass and root mass were measured in ancillary field experiments at Baiyun Experimental station in 2008 and 2011 to avoid the major disturbance of taking soil monoliths in the LTE. These ancillary experiments were near the LTE, on the same soil profile, and with largely similar management, with minor variations as described below. Both fields had a history of at least 20 years of wheat/maize intercropping in a small rotation, similar as in the LTE. In 2008, a completely randomized block design was used with three replicates and five treatments. The experimental treatments included sole crops of maize, wheat and faba bean and two intercrops: maize/wheat and maize/faba bean in a replacement design. Plot size was 31.5 m2 for maize/wheat intercrop (4.5 m width by 7 m row length) and 25.2 m2 for maize/faba bean intercrop (3.6 m width by 7 m row length). Management (fertilization, irrigation, tillage, etc.) was similar to the LTE, with two differences: (i) Each plot received 75 kg ha1 P as triple superphosphate (rather than 40), and (ii) the maize/wheat intercrop consisted of two maize rows at 40 cm distance, with 25 cm to the nearest wheat row, and six wheat rows at 12 cm distance. The maize/faba bean intercrop consisted of an 80 cm wide maize strip with two rows at 40 cm row distance, and a 40 cm wide faba bean strip with two faba bean rows at 20 cm distance. Sole crops of maize, wheat and faba bean were sown at row distances of 40 cm, 12 cm and 20 cm, respectively. Relative densities of maize and wheat in the intercrop were accordingly 0.53 and 0.48 respectively (i.e. a very small departure from pure replacement for which the sum of relative densities should be exactly one. Relative densities of maize and faba bean in the intercrop were 0.67 and 0.33, respectively, i.e. replacement. Harvested areas in sole crops were 5.6 m2 in maize (2 rows at 40 cm 9 7 m row length), 5.04 m2 in wheat (6 rows at 12 cm 9 7 m row length) and 2.8 m2 in faba bean (2 rows at 20 cm 9 7 m row length). The same areas were harvested in the intercrops. The experiment was about 400 m southwest of the LTE. Peak root biomass was determined in each plot from monoliths (B€ ohm, 1979) taken at the time of maximum root biomass: mid-June in wheat (milk stage) and faba bean (flowering

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

4 W - F . C O N G et al. covering 5 strips of both species in the intercrop plots. The field experiment was laid out as a completely randomized block design with three replicates. All management such as tillage, fertilizer and irrigation was applied uniformly across all treatments to avoid biasing the results. Each plot received 225 kg ha1 root biomass N as urea and 40 kg ha1 P as triple superphosphate annually with twothirds of the amount applied at moldboard tillage (down to 20 cm depth) just before sowing of wheat and faba bean in late March. The remainder was top-dressed at wheat flowering in late May. Irrigation was given seven times per year, five times during the growth periods of all three crops, and two times after harvest of wheat and faba bean (Fig. 2). Aboveground crop residues were removed after harvest for use as feed or fuel. Prior to sowing in 2003, a bulk soil sample was collected by taking cores 0–20 cm deep across the field. On 1 July 2010, soil samples were taken in each plot, when wheat was at the dough stage (White & Edwards, 2008), maize was in stage V10 (Abendroth et al., 2011), and faba bean was in pod-filling stage (Knott, 1990). Cores were taken using an auger (35 mm diameter) down to 1 m depth and subdivided in 0–20, 20–40, 40–60, 60–80 and 80–100 cm. Three cores per plot were taken in sole crop plots and bulked per depth. In intercrop plots, three subsamples of three cores each were taken in each plot, two from the crop strips, and a third one from the transition zone. Samples in the crop strip were taken in the middle between the maize rows in maize, in the middle between the 2nd and 3rd row in the faba bean strip, and in the middle between the 3rd and 4th row in the wheat strip. Soil Organic Carbon (SOC) and Soil Organic Nitrogen (SON) contents at the three locations within each intercrop plot were averaged after initial analysis demonstrated no significant differences between the three locations per plot. Soil samples were transported to the lab, air-dried for 3 days, and sieved over a 1 mm mesh sieve to remove plant debris. Samples were treated with 0.5 M HCl (10 ml g1 soil) to remove inorganic C (Midwood & Boutton, 1998). After drying, the acid-treated soil samples were ground in a ball mill. Acid-washing removes mineral N from the soil sample. Therefore, the analysis method results in quantification of organic N in the soil. SOC and SON were determined in duplicate using C/N analyzers in two chemical laboratories: C/N analyzer (Vario Macro, Elementar, Hanau, Germany) at the Stable Isotope Lab of China Agricultural University, and C/N isotope ratio spectrometer (PDZ Europa Integra, Cheshire, United Kingdom) at the Stable Isotope Facility of the University of California. Soil d15N was determined in the latter. The results on SOC and SON contents were pooled for data analysis, because systematic differences between the two determinations were not found and pooling allowed for greater accuracy. Part of the results on SOC (0–20 cm depth) were further validated against results from wet chemical oxidation (Kurmies) at Wageningen University which confirmed the results from the C/N analysers (cf. Walinga et al., 1992). To measure soil bulk density, six pits were dug, two in each of the three blocks of the experiment, and one in each crop system. Soil samples were collected by gently pushing 5 cm

diameter cylinders (length: 5.1 cm; volume: 100 cm3) laterally into the side of the pit at depths of 10 and 30 cm. Soil samples were oven-dried (105 °C) for 3 days before weighing. Soil organic C, Soil organic N, C/N ratio and d15N were statistically analysed using four-way repeated-measures ANOVA in SPSS 20.0 (IBM Corporation, Armonk, NY, USA) with depth as within-subject factor, and block, crop system (sole crops vs. intercrops) and crop combination (maize/wheat, maize/faba bean or wheat/faba bean) as between-subjects factors. SOC, SON and C/N ratio were further analysed separately for each soil layer using three-way ANOVA, with block as a random factor, and crop system and crop combination as fixed factors.

Single-year experiments on above- and belowground biomass Aboveground biomass and root mass were measured in ancillary field experiments at Baiyun Experimental station in 2008 and 2011 to avoid the major disturbance of taking soil monoliths in the LTE. These ancillary experiments were near the LTE, on the same soil profile, and with largely similar management, with minor variations as described below. Both fields had a history of at least 20 years of wheat/maize intercropping in a small rotation, similar as in the LTE. In 2008, a completely randomized block design was used with three replicates and five treatments. The experimental treatments included sole crops of maize, wheat and faba bean and two intercrops: maize/wheat and maize/faba bean in a replacement design. Plot size was 31.5 m2 for maize/wheat intercrop (4.5 m width by 7 m row length) and 25.2 m2 for maize/faba bean intercrop (3.6 m width by 7 m row length). Management (fertilization, irrigation, tillage, etc.) was similar to the LTE, with two differences: (i) Each plot received 75 kg ha1 P as triple superphosphate (rather than 40), and (ii) the maize/wheat intercrop consisted of two maize rows at 40 cm distance, with 25 cm to the nearest wheat row, and six wheat rows at 12 cm distance. The maize/faba bean intercrop consisted of an 80 cm wide maize strip with two rows at 40 cm row distance, and a 40 cm wide faba bean strip with two faba bean rows at 20 cm distance. Sole crops of maize, wheat and faba bean were sown at row distances of 40 cm, 12 cm and 20 cm, respectively. Relative densities of maize and wheat in the intercrop were accordingly 0.53 and 0.48 respectively (i.e. a very small departure from pure replacement for which the sum of relative densities should be exactly one. Relative densities of maize and faba bean in the intercrop were 0.67 and 0.33, respectively, i.e. replacement. Harvested areas in sole crops were 5.6 m2 in maize (2 rows at 40 cm 9 7 m row length), 5.04 m2 in wheat (6 rows at 12 cm 9 7 m row length) and 2.8 m2 in faba bean (2 rows at 20 cm 9 7 m row length). The same areas were harvested in the intercrops. The experiment was about 400 m southwest of the LTE. Peak root biomass was determined in each plot from monoliths (B€ ohm, 1979) taken at the time of maximum root biomass: mid-June in wheat (milk stage) and faba bean (flowering

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

0.06 0.49 0.72 0.43 0.93 0.55 0.32 0.29 0.30 0.78 0.61 0.80 0.75 0.07 0.15 0.35 0.07 0.88 0.09 0.09 0.03 0.16 0.58 0.40 0.94 0.00 0.00 0.01 0.39 0.40 0.15 0.17 0.21 0.17 0.20 0.01 0.02 0.03 0.03 0.03









12.50 8.81 6.62 5.88 5.34 1.47 1.10 0.89 0.79 0.73 0.13 0.15 0.20 0.10 0.40 0.02 0.02 0.02 0.02 0.05









12.05 8.43 6.41 6.05 5.31 1.36 0.98 0.77 0.76 0.69 0.29 0.21 0.58 0.17 0.38 0.02 0.02 0.07 0.04 0.06









12.04 8.72 6.79 6.35 5.88 1.49 1.11 0.93 0.87 0.81 0.21 0.36 0.35 0.29 1.09 0.01 0.01 0.05 0.03 0.14









12.25 8.70 6.34 6.09 5.84 1.39 1.04 0.74 0.77 0.76 0.18 0.33 0.35 0.21 0.37 0.03 0.03 0.06 0.04 0.05









12.61 9.04 6.51 5.34 5.00 1.46 1.12 0.91 0.73 0.69 0.31 0.10 0.50 0.15 0.67 0.06 0.01 0.04 0.00 0.07









12.05 8.27 6.15 5.93 4.80 1.38 0.96 0.77 0.72 0.61 0.09 0.41 0.31 0.27 0.10 0.01 0.03 0.03 0.05 0.01









SON

Significant effects (P < 0.05) are shown in bold. *Cs: crop system (sole crops vs. intercrops). †Cc: crop species combination.

12.86 8.68 6.55 5.96 5.14 1.47 1.06 0.83 0.76 0.68 0.20 0.35 0.31 0.18 0.50 0.02 0.05 0.03 0.04 0.06









11.85 8.32 6.75 6.11 5.29 1.31 0.93 0.80 0.79 0.68 0–20 20–40 40–60 60–80 80–100 0–20 20–40 40–60 60–80 80–100

0.39 0.53 0.75 0.40 0.79 0.05 0.05 0.08 0.07 0.10

Cs 9Cc Cc† Cs* Intercrops

P values Across crop combinations

Sole crops Intercrops Sole crops Intercrops Sole crops Intercrops

SOC

Soil organic C content differed significantly (P = 0.03) between sole and strip intercrop systems after 7 years. C content in the top 20 cm of the soil profile was 12.1
0.13 in sole crop systems vs. 12.5
0.15 g kg1 in intercrop systems (i.e. a difference of 4%) (Fig. 3a and Table 1), and 10.2
0.12 vs. 10.7
0.11 g kg1 (a difference of 3%) averaged over the top 40 cm (P = 0.04; Table S1). From 40–100 cm depth, SOC was similar in intercrop and sole crop systems. Soil organic N content differed by 8% in the top 20 cm (1.36
0.02 vs. 1.47
0.01 g kg1; Fig. 3b and Table 1) and by 11%

Sole crops

Soil C and N content

SED

Results

Depth (cm)

where Y1 is biomass density or biomass C, or N (g m2) of species 1 in intercrop, and M1 its biomass density or biomass C, N (g m2) in sole crop, and the same for species 2. Land equivalent ratio expresses the relative land area that would be required as sole crops to achieve the same yield or biomass as a unit area of intercrop. Land equivalent ratio is agronomically the most relevant measure, because it quantifies the production possibilities of the system taking into account that different crop species produce a different kind of biomass. When the LER is greater than 1, there is a land use advantage of intercropping (Vandermeer, 1989). An LER greater than 1 does not imply that the intercrop has a greater total yield or biomass per unit area than both sole crops (i.e. transgressive overyielding; Trenbath, 1974). We calculated LER for total aboveground biomass, root biomass and root biomass C and N.

Variables

Y1 Y2 þ M1 M2

Maize and faba bean

LER ¼

Wheat and faba bean

where RD1 + RD2 = 1 as all experiments were carried out as replacement designs. We compared the differences between observed and expected for total aboveground biomass, root biomass and root biomass C and N. Total aboveground and belowground biomass as well as total root biomass C and N over the 1 m profile were analysed using three-way repeated-measures ANOVA, with year as within-subject factor, and intercropping (expected vs. observed) and block as fixed factors. Two-way ANOVA was used to analyse the effects of intercropping and crop species combination on root biomass, root biomass C and root biomass N in each year as the first analysis indicated significant year 9 intercropping interaction. Data on root biomass C and N at five depths were analysed using four-way repeated-measures ANOVA with depth as within-subject factor, and block, intercropping and crop combination as between-subjects factors. Root biomass C and N at each depth were analysed using threeway ANOVA with block as a random factor, and intercropping and crop combination as fixed factors. To assess the agronomic efficiency of the tested intercrops, the land equivalent ratio (LER) was calculated:

Maize and wheat

Yexp ¼ Yexp;1 þ Yexp;2 ¼ RD1 M1 þ RD2 M2

Table 1 Soil organic C (SOC) and soil organic N (SON) (g kg1) in five soil layers in six crop systems, comprising three rotated strip intercrop systems (small rotations) and three rotations of sole crops (mean
SEM, N = 3). P values are from three-way ANOVA with block as random factor, and crop system (sole crops vs. intercrops) and crop species combination as fixed factors

6 W - F . C O N G et al.

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

INTERCROPPING ENHANCES SOIL C AND N 7 in the upper 60 cm (Table S1), i.e. the effect of strip intercropping on SON was more than two times as large as that on SOC. All effects of intercropping on SOC and SON were independent of which two crop species were combined (Table 1 and Table S1). Soil

Fig. 4 d15N over the soil profile (0–100 cm) in six crop systems after 7 years in the long-term experiment. Data are means
SEM, N = 3. Means with the same letter do not differ significantly within a crop combination (P = 0.05).

C/N ratio across the top 60 cm of the soil profile was significantly decreased by intercropping (Fig. 3c).

Estimation of soil C and N sequestration Soil C and N sequestration can be inferred directly from comparing C and N content in 2003 and 2010, while further evidence for sequestration potential of intercropping is obtained from quantifying the divergence in C and N stocks between the sole and intercrop systems as measured in 2010. Given an initial C content in the top 20 cm of 11.4 g kg1 in 2003, soils of both the sole crop and intercrop systems showed increases in C content over time; by 5.5% and 9.5%, respectively, averaged over crop species combinations; however, the increase was greater in the intercrop systems. Sole and intercrop systems also increased N content over 7 years; by 6.3% and 14.8%, respectively. Combining these changes in C and N content with the measured soil bulk density of 1.44 g cm3 in 2010 (equal in sole crop and intercrop systems; Figure S1), we estimate that C stocks in the top 20 cm of the soil in sole crop and intercrop systems diverged at a rate of 184
86 kg C ha1 yr1 from 2003 to 2010. The N

Fig. 5 Comparison of above- and belowground biomass in maize, wheat and faba bean and strip intercrops of maize with wheat or faba bean. For intercrops, the expected biomass is shown alongside observed biomass. Expected biomass is calculated as the weighted mean biomass in the sole crops, with weighing according to the relative density (intercrop/sole crop) of each species in the intercrop. Data are means
SEM (N = 3 in 2008; N = 5 in 2011). Asterisks refer to significant differences between expected and observed biomass. ***P < 0.001; **P < 0.01; *P < 0.05. Aboveground biomass in 2008 was previously reported (Li et al., 2011c). © 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

0.06 0.49 0.72 0.43 0.93 0.55 0.32 0.29 0.30 0.78 0.61 0.80 0.75 0.07 0.15 0.35 0.07 0.88 0.09 0.09 0.03 0.16 0.58 0.40 0.94 0.00 0.00 0.01 0.39 0.40 0.15 0.17 0.21 0.17 0.20 0.01 0.02 0.03 0.03 0.03









12.50 8.81 6.62 5.88 5.34 1.47 1.10 0.89 0.79 0.73 0.13 0.15 0.20 0.10 0.40 0.02 0.02 0.02 0.02 0.05









12.05 8.43 6.41 6.05 5.31 1.36 0.98 0.77 0.76 0.69 0.29 0.21 0.58 0.17 0.38 0.02 0.02 0.07 0.04 0.06









12.04 8.72 6.79 6.35 5.88 1.49 1.11 0.93 0.87 0.81 0.21 0.36 0.35 0.29 1.09 0.01 0.01 0.05 0.03 0.14









12.25 8.70 6.34 6.09 5.84 1.39 1.04 0.74 0.77 0.76 0.18 0.33 0.35 0.21 0.37 0.03 0.03 0.06 0.04 0.05









12.61 9.04 6.51 5.34 5.00 1.46 1.12 0.91 0.73 0.69 0.31 0.10 0.50 0.15 0.67 0.06 0.01 0.04 0.00 0.07









12.05 8.27 6.15 5.93 4.80 1.38 0.96 0.77 0.72 0.61 0.09 0.41 0.31 0.27 0.10 0.01 0.03 0.03 0.05 0.01









SON

Significant effects (P < 0.05) are shown in bold. *Cs: crop system (sole crops vs. intercrops). †Cc: crop species combination.

12.86 8.68 6.55 5.96 5.14 1.47 1.06 0.83 0.76 0.68 0.20 0.35 0.31 0.18 0.50 0.02 0.05 0.03 0.04 0.06









11.85 8.32 6.75 6.11 5.29 1.31 0.93 0.80 0.79 0.68 0–20 20–40 40–60 60–80 80–100 0–20 20–40 40–60 60–80 80–100

0.39 0.53 0.75 0.40 0.79 0.05 0.05 0.08 0.07 0.10

Cs 9Cc Cc† Cs* Intercrops

P values Across crop combinations

Sole crops Intercrops Sole crops Intercrops Sole crops Intercrops

SOC

Soil organic C content differed significantly (P = 0.03) between sole and strip intercrop systems after 7 years. C content in the top 20 cm of the soil profile was 12.1
0.13 in sole crop systems vs. 12.5
0.15 g kg1 in intercrop systems (i.e. a difference of 4%) (Fig. 3a and Table 1), and 10.2
0.12 vs. 10.7
0.11 g kg1 (a difference of 3%) averaged over the top 40 cm (P = 0.04; Table S1). From 40–100 cm depth, SOC was similar in intercrop and sole crop systems. Soil organic N content differed by 8% in the top 20 cm (1.36
0.02 vs. 1.47
0.01 g kg1; Fig. 3b and Table 1) and by 11%

Sole crops

Soil C and N content

SED

Results

Depth (cm)

where Y1 is biomass density or biomass C, or N (g m2) of species 1 in intercrop, and M1 its biomass density or biomass C, N (g m2) in sole crop, and the same for species 2. Land equivalent ratio expresses the relative land area that would be required as sole crops to achieve the same yield or biomass as a unit area of intercrop. Land equivalent ratio is agronomically the most relevant measure, because it quantifies the production possibilities of the system taking into account that different crop species produce a different kind of biomass. When the LER is greater than 1, there is a land use advantage of intercropping (Vandermeer, 1989). An LER greater than 1 does not imply that the intercrop has a greater total yield or biomass per unit area than both sole crops (i.e. transgressive overyielding; Trenbath, 1974). We calculated LER for total aboveground biomass, root biomass and root biomass C and N.

Variables

Y1 Y2 þ M1 M2

Maize and faba bean

LER ¼

Wheat and faba bean

where RD1 + RD2 = 1 as all experiments were carried out as replacement designs. We compared the differences between observed and expected for total aboveground biomass, root biomass and root biomass C and N. Total aboveground and belowground biomass as well as total root biomass C and N over the 1 m profile were analysed using three-way repeated-measures ANOVA, with year as within-subject factor, and intercropping (expected vs. observed) and block as fixed factors. Two-way ANOVA was used to analyse the effects of intercropping and crop species combination on root biomass, root biomass C and root biomass N in each year as the first analysis indicated significant year 9 intercropping interaction. Data on root biomass C and N at five depths were analysed using four-way repeated-measures ANOVA with depth as within-subject factor, and block, intercropping and crop combination as between-subjects factors. Root biomass C and N at each depth were analysed using threeway ANOVA with block as a random factor, and intercropping and crop combination as fixed factors. To assess the agronomic efficiency of the tested intercrops, the land equivalent ratio (LER) was calculated:

Maize and wheat

Yexp ¼ Yexp;1 þ Yexp;2 ¼ RD1 M1 þ RD2 M2

Table 1 Soil organic C (SOC) and soil organic N (SON) (g kg1) in five soil layers in six crop systems, comprising three rotated strip intercrop systems (small rotations) and three rotations of sole crops (mean
SEM, N = 3). P values are from three-way ANOVA with block as random factor, and crop system (sole crops vs. intercrops) and crop species combination as fixed factors

6 W - F . C O N G et al.

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

INTERCROPPING ENHANCES SOIL C AND N 9 4%), and 10.2
0.12 vs. 10.7
0.11 g kg1 averaged over the top 40 cm (a difference of 3%). We estimate that C stocks in the top 20 cm of the soil in sole crop and intercrop systems diverged at a rate of 184
86 kg C ha1 yr1 from 2003 to 2010. Furthermore, we demonstrated in ancillary experiments that root mass in strip intercrop systems is substantially greater (on average by 23%) than would be expected if the root mass per plant was the same as in sole crop systems. Over time, crop systems based on intercrops sequestered not only more soil organic C but also more soil organic N than crop systems based on sole crops. Intercrop systems had 11%
1% greater N stock in the top 20 cm of the soil after 7 years, equivalent to a divergence between in N stocks between intercrops and sole crops of 45
10 kg N ha1 yr1. The first plausible causal mechanism for the increase in SON is stoichiometry: proportional to the increase in belowground C input, more N was found in roots. The enhanced levels of organic soil N may be attributed to greater C input, which will immobilize some N, in combination with complementary N acquisition strategies of the intercropped species. Increase in soil N in the intercrops as compared to sole crops did not depend on the presence of legumes, but was in the studied systems supported by the input of fertilizer. These findings show that positive effects of biodiversity on ecosystem functioning observed in perennial N-limited grasslands with or without legumes (Fornara & Tilman, 2008; Cong et al., 2014) occur also in annual N-rich intercrops from which most of the biomass produced is removed at harvest, regardless of the presence of legumes. These here-to-fore unrecognized soil ecosystem services amplify well-known advantages of mixed crop systems in terms of productivity, production risk mitigation and suppression of pests and diseases (Vandermeer, 1989; Trenbath, 1993; Lithourgidis et al., 2011). The simultaneous increases in soil C and N suggest the possibility of synergy between C and N sequestration, whereby C sequestration enhances N sequestration, and vice versa. The mechanism for such synergy is a positive feedback loop whereby part of the N retained in organic matter would be remobilized yearly, contributing to greater aboveground and belowground productivity, which could result in further enhanced C sequestration and N capture, as well as retention. Several studies in grasslands found N-induced increases in total root mass in plots receiving over 20 years of N addition (Lu et al., 2011; Fornara & Tilman, 2012; Fornara et al., 2013). Such a synergy between C and N sequestration was also indicated by increasing strength of diversity effects over time in a grassland experiment without fertilizers or legumes (Cong et al., 2014). All in all, results of this and other studies indicate a linkage

between C sequestration potential of diversified plant systems, and the ability to retain N, whereby improved N capture and retention would increase productivity, in part due to yearly remobilization of some of the retained N. This linkage may result in a plant diversitydriven positive feedback between resource capture by the plant community and storage of C and N in soil over time, eventually resulting in a strengthening of diversity effects over time (Cardinale et al., 2012; Cong et al., 2014). If diversity-driven positive feedback effects between soil C and soil N occurred also in the long-term trial, then the measured effects of intercropping on root mass in the single-year experiments may be underestimates of the true intercropping effect on below ground productivity that would accrue over time. Further studies in a long-term experiment are needed to test whether plant responses increase over time as they have been found to do in grassland experiments (Cong et al., 2014). The finding that intercrops have greater root mass than expected from the equivalent sole crops is consistent with previous studies that demonstrated increases in root length density (Li et al., 2006; Gao et al., 2010). Also, for studies on root traits such as total root length per plant or per unit area of crop or per unit volume of soil, the question could be asked what root responses might occur if intercropping is maintained over time. Usually, intercropping studies are not conducted in a rotation context, but the increases of soil C and N over time indicate that the full response of the plants to intercropping may not be expressed until the soil C and N pools have developed their full response, which will not happen in 1 year. In addition, soil C/N ratio was lower in intercrops than in sole crops, suggesting that intercropping may increase the relative decomposition rate of soil organic matter (Booth et al., 2005). It has been long known that many intercropping systems used in the world by farmers have a land equivalent ratio above one (Vandermeer, 1989; Lithourgidis et al., 2011). A land equivalent ratio greater than one does not necessarily mean that in any 1 year, an intercrop would produce more biomass than both of the sole crops (i.e. transgressive overyielding) (Trenbath, 1974). Still, if a farm plan or rotation with (mostly) intercrops were compared to a farm plan or rotation with mostly sole crops, then land use systems with a greater proportion of intercrops (provided LER > 1) are expected to produce more biomass on the whole. When averaged out over fields on the farm or for a single field over time, an LER greater than 1 would result in overall greater biomass or yield, and in the case of soil carbon balance: greater C input. The finding of an association between aboveground and belowground overyielding

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

10 W - F . C O N G et al. in this study, and the finding of a linkage between belowground overyielding in intercrops and increased soil C and N levels suggests therefore that an increased use of intercropping could enhance the overall capacity of agricultural land to sequester C and N in soil. This study concerns only strip intercrop systems, which are common in Asia but less common elsewhere (Li et al., 2013). Strip intercrop systems have ‘inner rows’ which are crop rows that are bordered at each side by rows of the same species (in the same strip), and ‘border rows’, which are crop rows that are bordered at one side by the other species (in a neighbouring strip). Strip intercrops allow therefore – on average – a lower level of interspecific interaction than fully mixed or row-intercrops. They also allow a lower level of intraspecific interaction – on average – than sole crops. Strip intercrops are therefore in terms of interspecific interaction strength intermediate between sole crops and fully mixed or row-intercrops. It may therefore be anticipated that other intercropping patterns, i.e. completely mixed or row-intercropping, may show at least similar increases in soil C and N as those found in this study, but this suggestion needs empirical testing. Intercrops that consist of species with differing growing periods can intercept more light than sole crops, resulting in greater biomass and yield (Zhang et al., 2008). This mechanism for temporal niche differentiation is likely to result in greater belowground C input as crop species maintain a functional balance between the shoot and the root system (Poorter et al., 2012). This mechanism is likely to operate in intercrops with maize, as maize has a much longer growing season than wheat or faba bean. The lower soil d15N in intercrop rotations with faba bean in our study indicates that increased biological N fixation contributed to the enhanced N storage (Fig. 4). Previous studies demonstrated that intercropping legumes with cereal crops enhances biological N fixation of legumes, even at high N fertilizer level (Li et al., 2009). Intercropping with legumes also enhances N acquisition by cereals, first because competition for soil mineral N is reduced (Jensen, 1996b), and secondly because some of the N fixed by legumes can be transferred to cereals through root exudation (Jensen, 1996a) or mineralization of legume roots after harvest of the legume, but before harvest of the cereal. A lowering of d15N would also occur if denitrification was reduced, as gaseous losses discriminate against the heavier 15N isotope. This possibility cannot be ruled out, and is even plausible, as intercropping with cereals tends to lower soil nitrate levels in intercrops with faba bean (Li et al., 2005). Further work is needed to quantify these losses.

One of the key findings is that N accumulation was similar in systems with and without faba bean. As there was no change in d15N in the maize/wheat intercrop, losses that do not discriminate against d15N must explain the enhanced N storage. The key mechanism is probably a reduction in N leaching, which is a major pathway for N loss in the North China Plain (Ju et al., 2009). Reduced leaching is partly a consequence of functional complementarity between crop species in the location and timing of N uptake across the soil profile: Due to greater root biomass, presence of active roots during a longer part of the growing season (Fig. 2) and greater root biomass in deeper layers in the maize/wheat intercrop (Fig. 6a,b), the combined root systems of the two intercropped species can intercept nitrate that might otherwise be lost from the profile (Table S2; Fig. 5). In addition, N recycling within the growing season is possibly enhanced: after the harvest of the early maturing crops (wheat and faba bean) their root litter starts decomposing, releasing mineral N into the soil. Maize roots can proliferate in the N rich top layers (Drew & Saker, 1975) that were previously colonized by wheat or faba bean roots, and prevent the mineralized N from being lost from the soil system. Our findings extend earlier indications for higher N retention in more diverse natural grasslands (Tilman et al., 1996, 1997; Hooper & Vitousek, 1997) to high N input agricultural systems. The potential benefits of strip intercropping for improving long-term soil fertility by enhanced C and N sequestration may be more important than the possible contribution to greenhouse gas mitigation. Agricultural management practises such as zero-tillage, cover crops and complex rotations with deeper roots can enhance the rate of C sequestration by 50 to 500 kg C ha1 yr1 (West & Post, 2002; Lal, 2004; McDaniel et al., 2014). The effect of strip intercropping on soil C storage is within this range. The contribution to greenhouse gas mitigation in China is, however, minor. Assuming that one-third of China’s cropland may be used for strip intercropping (Zhang & Li, 2003), the C sequestration potential of strip intercropping in China is estimated to be 7.4 Tg C yr1, which accounts for only 0.4% of the total greenhouse gas emissions (6100 Tg CO2-eq) in China in 2004 (Zhang et al., 2013). Biodiversity effects on soil C and N storage in natural grasslands are associated with functional complementarity between plant species, e.g. between legumes that can fix N, and C4 grasses that can use this N efficiently to produce biomass (Fornara & Tilman, 2008). In our study, not only did we find functional complementarity between faba bean (legume, early growing season) and maize (C4, late growing season) but we also found functional complementarity between wheat (C3, early

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

8 W - F . C O N G et al.

(a)

(b)

(c)

(d)

Fig. 6 Standing root biomass C (a, c) and N (b, d) in five soil layers in maize/wheat (a, b) and maize/faba bean (c, d) strip intercrops in 2011 vs. expected values, calculated on the basis of root biomass C and N in sole crops (cf. Fig. 5). Data are means
SEM, N = 5. Asterisks refer to significant differences between expected and observed values at each depth. ***P < 0.001; **P < 0.01; *P < 0.05.

stocks in the top 20 cm soil profile diverged by 45
10 kg N ha1 yr1.

d15N Soil d15N was lower in intercrop systems with faba bean than in the maize/wheat intercrop (P = 0.001); it was also lower in the intercrop with faba bean than in the corresponding sole crop systems (P = 0.003; Fig. 4). As biological N fixation by legumes is associated with a lowering of the d15N signature of the soil (H€ ogberg, 1997), this result indicates greater biological N fixation by faba bean in intercrops than in sole crops. There is no d15N signal of biological N fixation in the sole faba bean.

Above- and belowground biomass In the single-year experiments, intercrops showed high land use efficiency for yield, characterized by land equivalent ratios for grain yield from 1.21 to 1.37 (Table S2). Furthermore, the aboveground biomass produced in intercrop systems was 13% to 23% greater than expected from sole crop biomass production (Fig. 5). Land equivalent ratio for belowground biomass ranged from 1.10 to 1.45 (Table S2). The maximum standing root biomass was on average 23% greater in intercrops

than expected from belowground biomass measured in sole crops (Fig. 5). Total root biomass C and N in intercrops exceeded the expected in both tested species combinations (maize/wheat and maize/faba bean) at all depths (Fig. 6). Significant differences were found at all depths except 40–60 cm in maize/wheat intercrop, whereas the difference between expected and observed root biomass C and N in maize/faba bean intercrop was only significant in the upper 20 cm. In the maize/faba bean, the increase in root biomass C and N in the topsoil could be attributed mainly to maize (Fig. 6c,d) whereas in maize/wheat intercrop there was also a contribution from wheat in the top 40 cm. Intercropping did not significantly affect root C and N concentrations in any of the three crop species (Figure S2). Root C/N ratios were 32 for maize, 30 for wheat and 22 for faba bean (significantly lower than in maize or wheat at P < 0.05).

Discussion Here, we show that strip intercrops accumulated higher levels of SOC and SON than sole crops over a period of 7 years. Carbon content in the top 20 cm of the soil profile was 12.1
0.13 in sole crop systems vs. 12.5
0.15 g kg1 in intercrop systems (a difference of

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738

12 W - F . C O N G et al. Lithourgidis AS, Dordas CA, Damalas CA, Vlachostergios DN (2011) Annual intercrops: an alternative pathway for sustainable agriculture. Australian Journal of Crop

Willey RW (1979) Intercropping: its importance and research needs. I. Competition and yield advantages. Field Crop Abstracts, 32, 1–10.

Science, 5, 396–410. Loreau M, Hector A (2001) Partitioning selection and complementarity in biodiversity experiments. Nature, 412, 72–76. Lu M, Zhou X, Luo Y et al. (2011) Minor stimulation of soil carbon storage by nitrogen addition: a meta-analysis. Agriculture Ecosystems and Environment, 140, 234–244. McDaniel MD, Tiemann LK, Grandy AS (2014) Does agricultural crop diversity

Yang CH, Chai Q, Huang GB (2010) Root distribution and yield responses of wheat/ maize intercropping to alternate irrigation in the arid areas of northwest China. Plant Soil and Environment, 56, 253–262. Zhang FS, Li L (2003) Using competitive and facilitative interactions in intercropping systems enhances crop productivity and nutrient-use efficiency. Plant and Soil, 248, 305–312. Zhang L, van der Werf W, Zhang SP, Li B, Spiertz JHJ (2007) Growth, yield and qual-

enhance soil microbial biomass and organic matteer dynamics? a meta-analysis. Ecological Applications, 24, 560–570. Mead R, Willey RW (1980) The concept of a land equivalent ratio and advantages in yields from intercropping. Experimental Agriculture, 16, 217–228. Midwood AJ, Boutton TW (1998) Soil carbonate decomposition by acid has little effect on d13C of organic matter. Soil Biology and Biochemistry, 30,

ity of wheat and cotton in relay strip intercropping systems. Field Crops Research, 103, 178–188. Zhang L, Van der Werf W, Bastiaans L, Zhang S, Li B, Spiertz JHJ (2008) Light interception and utilization in relay intercrops of wheat and cotton. Field Crop Research, 107, 29–42. Zhang WF, Dou ZX, He P et al. (2013) New technologies reduce greenhouse gas emis-

1301–1307. Myaka FM, Sakala WD, Adu-Gyamfi JJ et al. (2006) Yields and accumulations of N and P in farmer-managed intercrops of maize-pigeonpea in semi-arid Africa. Plant and Soil, 285, 207–220. Peters RD, Sturz AV, Carter MR, Sanderson JB (2003) Developing disease-suppressive soils through crop rotation and tillage management practices. Soil and Tillage Research, 72, 181–192.

sions from nitrogenous fertilizer in China. Proceedings of the National Academy of Sciences of the United States of America, 110, 8375–8380. Zhu YY, Chen HR, Fan JH et al. (2000) Genetic diversity and disease control in rice. Nature, 406, 718–722. Zhu JQ, Vos J, Van der Werf W, Van der Putten PEL, Evers JB (2014) Early competition shapes maize whole plant development in mixed stands. Journal of Experimental Botany, 65, 641–653.

Poorter H, Niklas KJ, Reich PB, Oleksyn J, Poot P, Mommer L (2012) Biomass allocation to leaves, stems and roots: meta-analyses of interspecific variation and environmental control. New Phytologist, 193, 30–50. Salvagiotti F, Cassman KG, Specht JE, Walters DT, Weiss A (2008) Nitrogen uptake, fixation and response to fertilizer N in soybeans: a review. Field Crops Research, 108, 1–13. Snapp SS, Blackie MJ, Gilbert RA, Bezner-Kerr R, Kanyama-Phiri GY (2010) Biodiversity can support a greener revolution in Africa. Proceedings of the National Academy of Sciences of the United States of America, 107, 20840–20845. Steinbeiss S, Bessler H, Engels C et al. (2008) Plant diversity positively affects shortterm soil carbon storage in experimental grasslands. Global Change Biology, 14, 2937–2949. Tilman D, Wedin D, Knops J (1996) Productivity and sustainability influenced by biodiversity in grassland ecosystems. Nature, 379, 718–720. Tilman D, Knops J, Wedin D, Reich P, Ritchie M, Siemann E (1997) The influence of functional diversity and composition on ecosystem processes. Science, 277, 1300–1302. Tilman D, Hill J, Lehman C (2006) Carbon-negative biofuels from low-input highdiversity grassland biomass. Science, 314, 1598–1600. Trenbath BR (1974) Biomass productivity of mixtures. Advances in Agronomy, 26, 177–210. Trenbath BR (1993) Intercropping for the management of pests and diseases. Field Crops Research, 34, 381–405. Vandermeer JH (1989) The Ecology of Intercropping. Cambridge University Press, Cambridge. Walinga I, Kithome M, Novozamsky I et al. (1992) Spectrophotometric determination of organic carbon in soil. Communications in Soil Science and Plant Analysis, 23, 1935–1944. Wardle DA, Bonner KI, Nicholson KS (1997) Biodiversity and plant litter: experimental evidence which does not support the view that enhanced species richness improves ecosystem function. Oikos, 79, 247–258. Weil RR, Magdoff F (2004) Significance of soil organic matter to soil quality and health. In: Soil Organic Matter in Sustainable Agriculture (eds Magdoff F, Weil RR), pp. 1–58. CRC Press, Boca Raton, FL, USA. West TO, Post WM (2002) Soil organic carbon sequestration rates by tillage and crop rotation. Soil Science Society of America Journal, 66, 1930–1946. White J, Edwards J (2008) Wheat Growth and Development. NSW Department of Primary Industries, State of New South Wales, Australia.

Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1. Soil bulk density at 0–20 and 20–40 cm depth in sole crop and strip intercrop plots. Data are means
SEM, N = 3. Means with the same letter do not differ significantly (P = 0.05). Figure S2. Root C (a) and N (b) concentrations of maize, wheat and faba bean in sole crops and maize/wheat and maize/faba bean strip intercrops. Data are means
SEM, N = 4. Means with the same letter do not differ significantly between treatments within the same crop (P = 0.05). Table S1. Average soil organic C (SOC) and organic N (SON) (g kg1) from the top of the soil profile to different depths in six crop systems comprising three rotated strip intercrop systems (small rotations) and three rotations of sole crops (means
SEM, N = 3). P values are from threeway ANOVA with block as random factor, and crop system (sole crops vs. intercrops) and crop species combination as fixed factors. Significant effects (P < 0.05) are shown in bold. Table S2. Land equivalent ratio (LER) of aboveground biomass, grain yield, standing root biomass, standing root biomass C and N in maize/wheat and maize/faba bean strip intercrops in 2008 and 2011.

© 2014 John Wiley & Sons Ltd, Global Change Biology, doi: 10.1111/gcb.12738