Soil organic carbon in relation to cultivation in arable and greenhouse cropping systems in Lanzhou, NW China

Abstract

Soil organic carbon (SOC) is a major source/sink in atmospheric carbon balances. Farmland usually has a high potential for carbon dioxide (CO₂) uptake from the atmosphere, but also for emission. Data from different areas are valuable for global SOC calculations and model development, and a survey of 108 agricultural fields in Lanzhou, China was performed. The fields were grouped by: cropping intensity (3 levels), cropping methodology (3), and crop species (10). Intensive cropping (two or more crops per year, typically vegetables), moderate (annuals in monoculture: wheat, maize, potato, melons), and extensive (orchards, lily [Lilium brownii] fields, fallow) were the intensity classes; and open field, greenhouse field, and sand-covered field (10–20 cm added on top of the topsoil) were the three methodologies. SOC concentration, pH, electrical conductivity, and soil bulk density were measured, and SOC mass (g·m⁻² 0–20 cm depth) was calculated. SOC concentration was high in cauliflower, wheat, leaf vegetables, and fruit vegetables; moderate in potato, fallow (3–5 years), tree orchards, and melons; while low in lily and maize fields, and differences in SOC mass followed the same pattern. SOC concentration and mass were lowest in the extensive fields while moderate and intensive fields showed higher values. Soil bulk density in open fields was significantly lower than those in greenhouse and sand-covered fields. The climate-induced soil activity factor r_{e_clim} was calculated, compared with European conditions, and was fairly similar to those in central Sweden. Other factors behind the measured results, such as the influence of initial SOC content, manure addition, crops, etc., are discussed.

Keywords:

• agriculture
• soil organic carbon
• Gansu province
• land use
• r_{e_clim}

Introduction

Soil organic matter levels are crucial for the physical, chemical, and biological properties of soil; and soil organic carbon (SOC) is the largest C sink in terrestrial ecosystems. Thus, changes in the SOC pool size are directly linked to changes in atmospheric carbon dioxide (CO₂) concentration. China has 122 million hectares of farmland, and arable soil carbon data from NW China are scarce.

At a given moment, the SOC level in an agricultural soil depends on preagricultural SOC content and the inputs and outputs after first cultivation. The balance is affected by a number of climatic, edaphic, and crop-related factors such as temperature, precipitation, potential evapotranspiration, topography, crop yield, root/shoot ratio, etc. Since the agricultural ecosystem by definition is managed, cultivation, fertilization, manure addition, and irrigation are major determinants (Andrén & Kätterer 2008).

Intensity of the management is a major factor in SOC dynamics in agricultural ecosystems, and manure addition and fertilization will increase crop yields and SOC in soil (Bolinder et al. 2012), whereas cultivation and irrigation often will increase SOC decomposition rates and decrease SOC mass, but not always (Angers & Eriksen-Hamel 2008). A productive crop, and even several harvests per year, will give more residues to the soil and also transpire more water, thereby reducing SOC decomposition rates.

The artificial environment of a plastic greenhouse has been welcomed in China due to the potential for radical increases in vegetable yield compared with open-field cultivation. During the last 20 years, greenhouse cultivation in China has increased gradually; currently the greenhouse area including simple plastic tunnels is around 3 million ha (exact area depending of definition of ‘greenhouse’) of which 2.5 million ha is in Northern China (Tian et al. 2011). Sand-covering of fields, i.e., sand mulching, is a special amendment in arid land in Northern China, common in the central Gansu province. Gravel is often used for the same purpose (Li 2002; Xie et al. 2006). This measure increases rain interception, is well proven since more than 300 years (Li et al. 2005), and the low capillarity of the sand also reduces evaporation and conserves water (Li & Liu 2003). The coincidence of summer rainfall, high temperatures, and abundant sunlight makes the region very suitable for production of vegetables, and the Lanzhou area has traditionally been named the “Plateau Summer Vegetable Factory.” Naturally, Lanzhou is situated on the Yellow River, which facilitates irrigation – a central feature in Chinese agriculture. Large amounts of high-quality vegetable products (in a wide sense) were and are supplying nearly all cities of China, and some species such as lily bulbs and Chinese cabbage are exported to Japan and South Korea.

In summary, there are a number of interesting features of Gansu and Lanzhou agriculture, but knowledge about the relations between agricultural land use and SOC dynamics in the Gansu province is scarce, especially regarding the “industrialized farmland” such as greenhouses and sand-covered fields. Therefore, the main objective of this study is to investigate SOC concentrations and pools under different cropping systems in Lanzhou, Gansu province and discuss possible factors explaining the results.

Material and methods

Study site and climate

This study was conducted in three counties and five districts in the Lanzhou city area, in the central Gansu province, NW China. The research area ranges from 102°30′ to 104°30′ E, and 35°5′ to 38°N, with a total area of about 13,000 km² (Figure 1). The landscape in Lanzhou is characterized by mountains up to 2000 m alternating with undulating valley bottoms (at about 1500 m altitude) with rivers of varying size, including the Yellow River. The Maxian Mountain is located in the southeast and the Jiangjunfu Mountain in the northwest. The Yellow River divides the city of Lanzhou (Urban population from 2010 census: 2.2 million); there are also other rivers used for agricultural irrigation, including the Huangshui River in the Honggu district running from west to east, and the Zhuanglang River in Yongdeng County running from north to south.

Figure 1. Location of the studied region with soil sampling sites indicated.

The soils in the region typically are silt loam of loess origin, Haplic Orthic Aridisols (Li & Liu 2003). The climate is temperate, of the continental semi-arid monsoon type with a long-term annual mean temperature of about 9.4°C. Annual mean precipitation (314 mm) has high seasonal variability, and annual mean potential evapotranspiration is as high as 1450 mm (Feng et al. 2004). The rainy season is from July to October, in synchrony with the warm season, which is very beneficial for high-value agricultural production such as vegetable cultivation.

The mineralization activity in soil can be quantified by activity factors based on daily soil water and temperature from meteorological data. We used daily standard meteorological data 1952–1988 (temperature, precipitation, pan evaporation) to calculate reference evaporation (Et0; Allen et al. 1998) and the daily soil climate/activity factor (r_{e_clim}), which gives the relative activity of soil organisms compared to that in central Sweden (set to 1 there). The r_{e_clim} factor is calculated from activity estimates based on a standard soil “moved” to the local climate, under which daily soil water and temperature are calculated. These variables are fed to functions converting soil water and temperature to activity factors, r_{e_temp} and r_{e_wat}, which are multiplied for each day: r_{e_clim =} r_{e_temp ×} r_{e_wat.} Then an annual mean can be calculated, taking the daily interaction between soil temperature and water content into account (Andrén et al. 2007; Bolinder et al. 2007). Various sums and means were calculated and presented below and the programming was made in R (R Core Team 2013).

Typical vegetables in the region include lily (Lilium brownii, bulbs used as food and in traditional medicine), cauliflower, potato, Chinese cabbage, tomato, melons, etc. Also grain crops (wheat and maize) and fruits (apple, pear, peach, etc.) are common.

Experimental design

In all, 108 fields were sampled, being selected from a local database on cropping systems. The fields were chosen to cover the variation in crops, cropping intensity, and cultivation strategies in the area. The results (SOC concentration and carbon mass) were then independently analyzed in three ways: according to crop species, intensity, and cropping strategy (greenhouse, free land, or sand-covered).

The crop species (term used in a wide sense) used as one grouping factor in the analyses are listed in Table 1. The second factor, intensity, was investigated grouping the soil data into three intensity levels: intensive cropping (two or more crops per year, typically vegetables: Chinese cabbage, cauliflower, pumpkin), moderate (annuals in monoculture: wheat, maize, potato, melons), and extensive (orchards, lily [L. brownii] fields growing 6–7 years before harvest, fallow fields). The classes were represented by 26, 69, and 13 plots from intensive, moderate, and extensive fields, respectively. The third grouping factor, cultivation strategy, was as follows: open field (93 fields), greenhouse (11), and sand-covered field (10–20 cm added on top of the topsoil, 4 fields in all).

Table 1. Number of fields sampled for each crop species.

Species	Sample size (n)	Proportion (%)	Strategy of land use
Cauliflower	5	4.6	Open/greenhouse
Fallow (3–5 years)	3	2.8	–
Fruit vegetables (cucumber, tomato, pumpkin)	7	6.5	Open/greenhouse
Leaf vegetables	12	11.1	Open/greenhouse
Lily	3	2.8	Sand-covered
Maize	28	25.9	Open
Melons	4	3.7	Sand-covered
Orchard crops	17	15.7	Open/greenhouse
Potato	5	4.6	Open
Wheat	24	22.2	Open
Total	108	100	–

Soil sampling and laboratory analysis

Soil (0–20 cm depth) was collected from 108 plots. In each plot, we selected a quadrat, 300 – 800 m², depending on field size. Five to seven samples were randomly collected with a shovel in each quadrat. The subsamples from one plot were bulked and thoroughly mixed and a 1 kg sample was taken to the laboratory for chemical and physical analysis. Soil bulk density was measured by a cutting ring (0–20 cm, 100 cm⁻³ vol.), five replicates in each plot. The sand-covered plots needed special attention; the sand layer was gently removed before sampling.

Soil pH was measured in a 1:2.5 (w/w) and electrical conductivity in a 1:5 (w/w) soil:water suspension at 25°C (Multiline F/SET-3; WTW, Weilheim, Germany); total SOC was measured using the Walkley–Black dichromate oxidation procedure (Nelson & Sommers 1982). This method used excludes carbonates and is considered to be a good measure of SOC concentration (Wang et al. 2012).

SOC mass was calculated as follows (Novara et al. 2013):

where SOCs is SOC mass; BD is soil bulk density; H is thickness of soil layer; SOC_c means SOC concentration; CF coarse is a correction factor, [1 - (gravel% + stone%)] / 100. However, in this study, gravel and stones were not present in these fields, so CF coarse = 1.

Statistical treatment

The statistical analysis of data was conducted using SPSS (SPSS 13.0). The results were separately analyzed by one-way analysis of variance (ANOVA) with the different species, intensity, and cultivation strategy as factors, respectively. Multiple comparisons using the least significant differences (LSD) test were done when the ANOVA indicated significant differences (p < 0.05).

Results

Climate and soil activity

The climate in Lanzhou is summarized in Figure 2, based on the high-quality data-set compiled by Feng et al. (2004). During 1952–1998 the annual rainfall varied between 200 and 509 mm, with a mean of 314 mm, and the distribution over the year was mainly in the warmer months. This resulted in a high-microbial activity in the soil in summer, with rapid mineralization of plant nutrients from soil organic matter as well as from added manure and other organic amendments.

Figure 2. A summary of the climate in Lanzhou and the relative activity in soil 1952–1988. The soil activity factor r_{e_clim} is shown both for Naiman and Uppsala, Sweden (temperature, precipitation, Et0 = daily air temperature °C, precipitation mm, reference evapotranspiration mm; Dtemp, Dprec, DEt0 = mean daily values for 1952–1998, Dday = Julian day number; Tempmean = mean annual temperature °C, Precsum = annual precipitation sum mm, Mprec = mean daily precipitation (mm) by month 1952–1998; Yrr = year, Mmonth = month; Dre or Dreu = mean daily r_{e_clim} climate index 1952–1998, Dday or Ddayu = Julian day number; reann = annual r_{e_clim} 1952–1998).

The 1952–1998 mean annual soil activity factor (r_{e_clim}) was 1.3 in Lanzhou, while the baseline value in central Sweden is 1 (Figure 2). Thus the relative decomposition activity in Lanzhou is 1.3 times that in central Sweden (or agricultural land in Canada), which means that 1.3 times the annual inputs are needed to maintain Swedish soil C levels, and that about 55% of incorporated cereal straw is lost as CO₂ the first year.

However, the annual rainfall can be erratic, and the activity factor was as low as 0.97 in 1982, mainly due to low rainfall in spring. For comparison, annual r_{e_clim} in West African rainforest zones can be as high as 5, resulting in low-SOC levels in spite of high-primary productivity (Andrén et al. 2007).

Comparing the daily soil activity factor between Lanzhou and Uppsala, Sweden (average annual precipitation 533 mm, mean temperature 5.6°C), the similarity may be surprising (Figure 2, bottom row). Lanzhou is about 34°N and 104°E, whereas Uppsala is 60°N and 17°E. However, both are on the Eurasian landmass and in the northern hemisphere, so the climate pattern should not be too different. Uppsala is also 40 m above sea level (asl) while Lanzhou is at about 1500 m asl. which reduces summer temperatures – which together with the comparatively low precipitation in Lanzhou explains the relatively small absolute difference – only a factor of 1.3.

Species

Soil bulk density under the different species ranged from 1.27 to 1.36 g·cm⁻³, but the apparent differences were not statistically significant (p = 0.12), and the same was true for soil pH (p = 0.06). There were no significant differences in electrical conductivity between different crop species (p = 0.98), but there was higher variability compared with those in pH and bulk density (Table 2). SOC concentrations differed considerably between crop species (F = 4.18, p = 0.0001). Due to the minor differences in bulk density, the significant differences in soil C mass were similar to those in SOC concentration (F = 4.52, p < 0.0001); high in cauliflower, wheat, and leaf vegetable fields and low in lily and maize fields.

Table 2. Average soil properties of agricultural fields in Lanzhou, grouped according to crop type (±standard error).

Species	Bulk density (g·cm^−3)	pH	Electrical conductivity (μS·cm^−1)	SOC conc. (g·kg^−1)	SOC mass (mg·ha^−1)
Cauliflower	1.28 ± 0.06	8.50 ± 0.11	0.28 ± 0.07	13.84 ± 1.82a	36.00 ± 5.91a
Fallow	1.28 ± 0.03	8.69 ± 0.02	0.38 ± 0.25	8.88 ± 1.96bcd	25.16 ± 6.01abc
Fruit vegetables	1.34 ± 0.05	8.56 ± 0.10	0.31 ± 0.08	11.42 ± 1.30abc	30.69 ± 3.21ab
Leaf vegetables	1.36 ± 0.04	8.39 ± 0.10	0.35 ± 0.06	11.89 ± 1.68ab	31.51 ± 3.91a
Lily	1.36 ± 0.03	8.67 ± 0.06	0.31 ± 0.11	6.73 ± 2.52d	18.25 ± 6.96c
Orchard	1.36 ± 0.02	8.64 ± 0.05	0.35 ± 0.11	8.77 ± 0.82cd	23.65 ± 2.07bc
Potato	1.27 ± 0.07	8.80 ± 0.06	0.21 ± 0.05	9.49 ± 2.06bcd	23.32 ± 3.86bc
Maize	1.28 ± 0.02	8.64 ± 0.06	0.34 ± 0.14	7.47 ± 0.68d	19.07 ± 1.65c
Melons	1.34 ± 0.06	8.66 ± 0.04	0.29 ± 0.07	8.67 ± 1.94bcd	22.76 ± 4.51bc
Wheat	1.29 ± 0.01	8.59 ± 0.05	0.48 ± 0.16	13.07 ± 0.79a	33.92 ± 2.21a
Average	1.31 ± 0.01	8.59 ± 0.03	0.35 ± 0.04	10.11 ± 0.43	26.49 ± 1.11

Note: For SOC, values with different characters differ significantly between crop type (p < 0.05).

Intensity of land use

Grouping the soil according to cultivation intensity did show small but significant differences in soil bulk density and SOC (Figure 3). Soil bulk density in the moderate fields was lower (p < 0.001) than in the other two intensities, probably due to less traffic in the fields. In the intensive group, traffic related to irrigation, fertilization, and harvests was more frequent compared with maize or wheat fields. In extensive fields with, e.g., fruit trees, operations involving traffic (i.e., cutting, fertilizing, irrigation, and pesticide application) were performed. Naturally, there was also less frequent soil cultivation in these extensive fields.

Figure 3. Soil bulk density (A), SOC concentration (B), and mass (C) under different cultivation intensities. Bars with different characters differ significantly (p < 0.05), and thin bars indicate standard error.

The significantly higher SOC mass (p < 0.05) in the intensive fields can be explained by the high productivity and crop residue inputs here. With more than one crop per year where the total root system and leaves and stems not exported were cultivated into soil, the inputs to soil were high on an annual basis.

A high but not quantified amount of manures were used in the intensive fields (mainly vegetables) and orchards (such as pear, apple, peach), but also in wheat fields. However, manure application was and is limited by accessibility.

Strategy of land use

The third grouping factor, cultivation strategy, was as follows: open field (93 fields), greenhouse (11), and sand-covered field (four fields in all, 10–20 cm sand added on top of the topsoil).

In open fields, soil bulk density was 1.30 g·cm⁻³, which was significantly lower (p = 0.004) than that in greenhouse soils while not significantly different from that under sand cover (see Figure 4). Naturally, the greenhouse soil sees more traffic due to intensive management and limited space, which may explain the observation. The low number of sand-covered fields (n = 4) makes the comparisons difficult, and the sand-covered fields also showed high variability due to partial mixing between the top sand layer and the actual topsoil.

Figure 4. Soil bulk density (A), SOC concentration (B), and mass (C) under different cultivation strategies. Open = traditional freeland agriculture, Greenhouse; and sand-covered = fields covered with a 20-cm sand layer. Bars with different characters differ significantly (p < 0.05), and thin bars indicate standard error.

The lack of significant differences in SOC may be explained by the fact that the same factors that increase crop yield and consequential inputs to soil also increase decomposition rates (irrigation, reducing transpiration, increasing soil temperature; see, e.g., Swift et al. 1979 and Andrén et al. 2007). Also, the sand cover will very much reduce the feedback of above-ground plant residues to soil, and SOC contents can be lower than in greenhouse or open fields due to this. Then again, although not investigated here, the sand layer will with time gain SOC due to root turnover, and this SOC mass should also be taken into account for a complete soil carbon budget.

Discussion

Topsoil SOC mass ranged between 18 and 36 mg ha⁻¹ (0–20 cm depth). There are a number of reasons for these relatively high values, both natural and related to cultivation measures. After the growing season (April–October), the temperature plummets down below 0°C and the SOC, including the current year's crop residues including roots, is effectively conserved during the winter (see the soil activity diagram in Figure 2 – the decomposer activity is virtually zero during winter). The decomposition rate of SOC a given day is determined by the temperature and water conditions which interact in a multiplicative manner (Andrén et al. 2007), and in Lanzhou the November–March period is both cold and dry.

As mentioned in the introduction, SOC in an agricultural soil depends on preagricultural SOC content and the inputs and outputs after first cultivation as well as the time passed since the first cultivation. The balance is affected by a number of climatic, edaphic, and crop-related factors such as temperature, precipitation, potential evapotranspiration, topography, crop yield, root/shoot ratio, etc. Since the agricultural ecosystem by definition is managed, cultivation, fertilization, manure addition, and irrigation are major determinants (Andrén & Kätterer 2008). It may take many years of cultivation before a stable SOC level is reached, particularly if the preagricultural SOC content was very different from the steady-state value under agriculture. The younger SOC fractions, such as current year crop residues, will naturally vary even within a year, but the older (humus) fractions will be less dynamic (see, e.g., Andrén & Kätterer, 2001; Álvaro-Fuentes et al. 2012).

Based on one sampling occasion, it is difficult to separate the effects of preagricultural SOC levels from those from the cultivation measures. If the initial land use choice, e.g., greenhouse or open cultivation, depended on SOC content in the field, the measured SOC content could partly be dependent on this choice. Naturally, as time passes, the effect of these initial conditions decreases. In the present case, most of the cropping systems have been in place for a long time, but the greenhouse areas have increased in recent years. The selection of land for greenhouse cultivation, however, was based more on access to water and transportation as well as security reasons (guarding the crop) than on initial soil properties. Therefore, we conclude that the measured SOC differences mainly reflect the cropping systems, but cannot entirely rule out bias due to initial selection of fields.

The differences in soil carbon mass between crop species grown (Table 2) are probably not due to the species per se, but a consequence of different levels of manure addition and other cultivation measures. As can be seen in Figure 3, the intensive cultivation category (mainly greenhouse crops) showed high soil carbon mass, and the greenhouse strategy (Figure 4) also showed high soil carbon mass.

It is well known that both manure and fertilizer addition will increase SOC content (Gami et al. 2009; Mazzoncini et al. 2011; Álvaro-Fuentes et al. 2012). In Lanzhou, manure application differs between the open field and the greenhouse. For example, in the autumn, manure is often added to the winter wheat field to maintain a relative high temperature in the root zone and reducing frost damage (which probably explains the high C content under wheat, cf. Table 2). Manure is also added before sowing, but the application of manure is limited by availability and transport facilities. Since the greenhouse is a major investment, the soil is amended by very large amounts of manure before the greenhouse is erected. For example, in Table 2 it can be seen that the greenhouse crops are grown in a high-carbon soil, most probably due to this treatment.

Maize is an important cash crop in Lanzhou, and since it responds well to chemical fertilizer (Sui et al. 2009), manure is seldom used for this crop. After harvest, maize stovers are usually exported from the field and used as fodder or firewood. Compared with maize, wheat crop residues are rarely exported. Exporting crop residues or leaving them in the field makes great differences in annual carbon input to the soil (Tian et al. 2011). It has also been observed that root-derived carbon contributes more to relatively stable soil C pools than the same amount of above-ground residue-derived C (Kätterer et al. 2011), and this can partly explain why the sand-covered soils did not have a significantly lower carbon content in spite of the lack of input from above-ground plant residues.

The results from the Lanzhou survey clearly indicate that SOC levels partly can be predicted from climate, crop species, cropping intensity, and cultivation strategy (greenhouse vs. open field, etc.). A full explanation and solid base for long-term soil C modeling would need more actual measures, e.g., manure addition, export of crop residues, etc. However, surveys of SOC content and climate-based calculations will increase the understanding of the underlying factors and serve as a foundation for further research.

Acknowledgments

We thank Professors S Feng, Q Hu, and W Qian for permission to use their high-quality climate data-set and help in obtaining the same. We also thank colleagues of the Naiman Desertification Research Station, Chinese Academy of Sciences, for their help in laboratory analyses. Programs (in R, SAS and Excel) for calculating the activity factor r_{e_clim} can be downloaded from www.oandren.com/ICBM.

Funding

This work was financially supported by the National Nature Science Foundation of China [grant number 41071185 and 31170413]; the National Basic Research Program [grant number 2011BAC07B02]. The Chinese Academy of Sciences has kindly granted Prof. O. Andrén a ‘Professorship for Senior International Scientists’ [grant number Y229D91001].

Additional information

Funding

Funding: This work was financially supported by the National Nature Science Foundation of China [grant number 41071185 and 31170413]; the National Basic Research Program [grant number 2011BAC07B02]. The Chinese Academy of Sciences has kindly granted Prof. O. Andrén a ‘Professorship for Senior International Scientists’ [grant number Y229D91001].