Soil respiration versus vegetation degradation under the influence of three grazing regimes in the Songnen Plain

2.1 Study area

The research site was at the Grassland Farming Research Station (44°33′N, 123°31′E) of the Chinese Academy of Sciences in Changling County (northeast China). The study area has a semiarid continental climate. Daily mean air temperatures and precipitation during the duration of the study (2012 and 2013) are shown (Figure 1). Total precipitation during the growing season (May–September) was 494 and 330 mm in 2012 and 2013, respectively, corresponding to 143 and 96%, respectively, of long-term averages in this region (Figure 1). Maximum monthly rainfall occurred in June 2012 (152 mm) and July 2013 (146 mm). The soil is classified as meadow Solonchaks (FAO/UNESCO taxonomy; Jiang, He, Wu, & Zhou, 2010). The main vegetation in the study area was meadow steppe dominated by L. chinensis (perennial C₃ plant), whereas in some patches with higher soil pH and less soil nutrients, the dominant species was C. virgata (annual C₄ plant; Jiang et al., 2010). Initial and pre-experiment (2011) vegetation and soil characteristics at this study site are shown (Table 1).

2.2 Experimental design

In April 2012, three adjacent sites (each 100 × 200 m) located in a meadow based on grazing history and vegetation cover in 2011 (Figure 2) were identified. Before 2009, Sites 1 and 2 had the same ground cover status with light vegetation degradation. Site 1 (light degradation [LD]) was fenced for complete livestock exclusion since 2009, resulting in very light vegetation degradation and >80% vegetation coverage when the experiment started. In contrast, Site 2 (MD) was continuously grazed by sheep since 2009, using a moderate stocking rate of 2 sheep·ha⁻¹. There was moderate vegetation degradation with 50–60% vegetation coverage when the experiment started. Site 3 (severe degradation [SD]) was separated from Sites 1 and 2 by a ditch and had been heavily grazed by sheep and cattle for 15 years, with >6 sheep equivalent·ha⁻¹. Consequently, there was severe vegetation degradation, with <30% vegetation coverage when the experiment started. Grazing regimes were maintained at each site throughout the experimental period. There were five sampling strips across three sites, dividing each site into five blocks, 40 × 100 m, with each block divided into two plots measuring 40 × 50 m (Figure 2). Two community patches dominated by L. chinensis and C. virgata were randomly located in two plots. In the center of each patch, two adjacent subplots (2 × 2 m) were established as measuring and sampling subplots, respectively.

2.3 Soil respiration measurement

At each measuring subplot, soil respiration was measured every 2 weeks from May to September in 2012 and 2013. To measure total soil respiration (SR_TOT), 1 day prior to each measurement, a new small quadrant (15 × 15 cm) was selected in each measuring subplot. Plants and litter were carefully removed from each quadrant, with care taken to minimizing soil surface disturbance. Soil microbial respiration (SR_H) was measured to separate root respiration (SR_A) from total soil CO₂ efflux. In April 2012, a poly-vinyl chloride tube (15-cm diameter and 73-cm length) was inserted 70 cm into the soil in the measuring subplot and was left there for the duration of the experiment. To exclude SR_A when SR_H was measured, clipping treatment was conducted to ensure no living plants appeared in these plots throughout the whole experimental period of 2 years. A CIRAS-2 photosynthesis system coupled with a portable, closed soil CO₂ flux chamber (PP-Systems, Hitchin, UK) was used to measure soil respiration in quadrants and tubes. Because of the low rates of soil respiration, each measurement required 120 s to ensure reliable data. To avoid influences of rainfall on soil CO₂ efflux, measurements were performed at least 3 days after the last rainfall. Soil respiration was measured between 07:00 and 10:00, which has been reported to be representative of the mean daily value in this grassland (Wang et al., 2015). Experimental plots were measured in random order to avoid biased estimates. Adjacent to each soil respiration measurement site, soil temperature (°C) at the 10-cm soil depth was measured using probes connected to an FDS-100 Automatic Temperature Recorder (Handan Electronic Technology Company, Handan, China). In addition, soil moisture was measured at 0–10 cm of soil depth by a soil core method (ISSCAS, 1978).

2.4 Vegetation sampling and analysis

In late August of 2012 and 2013, aboveground biomass (AGB) was harvested by clipping a 0.5 × 0.5 m quadrant in each sampling subplot. Furthermore, all plant litter was collected in each quadrant. The belowground net primary productivity (BNPP) was estimated by the root ingrowth core method (Yan et al., 2010). In late April 2012 and 2013, a soil core (10 cm in diameter and 30 cm deep) was sampled and divided into three 10-cm-deep sections in each sampling subplot. After visible root materials were removed from the bulk soil of each section, root-free soil samples were replaced into the hole in the order that they were originally distributed. In late September, central soil samples of ingrowth cores were collected using an auger (7 cm in diameter) at three depths (0–10, 10–20, and 20–30 cm). Root materials were carefully removed from soil samples and collected. All plant materials were oven-dried at 65 °C for 48 hr, and their dry weight was determined.

2.5 Soil property determination

In 2012 and 2013, after sampling aboveground vegetation, three soil samples were collected using an auger (5 cm in diameter) and then combined into a composite sample at 0–10 cm of soil depth in each quadrant. With visible plant materials and other debris removed, soil samples were sieved through a 2-mm screen and then stored at 0 °C to determine soil microbial biomass carbon (MBC). In 2013, an additional composite sample was collected at depth of 0–10 cm in each quadrant. With similar preprocessing as above, these soil samples were air-dried in the dark and chemical properties were subsequently determined. In each quadrant in 2013, soil bulk density at 0–10 cm of soil depth was determined using the core method (Klute, 1986). Soil MBC was measured using the fumigation–extraction method (Liu, Zhang, & Wan, 2009). Soil pH was measured using a PHS-3C pH meter in a 1:5 soil water solution. Furthermore, SOC, soil total nitrogen (TN), and soil total phosphorus (TP) concentrations were determined on the basis of a physical and chemical analysis protocol (ISSCAS, 1978).

2.6 Temperature sensitivity of soil respiration

Relationships between soil respiration, soil temperature, and soil moisture were established on the basis of an exponential function when soil moisture had no significant influence on soil respiration (Equation 1) and a linear function when soil temperature and soil moisture jointly influenced soil respiration (Equation 2; Zhang et al., 2014).

(1)

(2)

where SR represents different soil respiration components (g·m⁻²·h⁻¹), LN(R) is the natural logarithm of SR, T represents soil temperature (°C), M represents soil moisture (%), and a, b, and c are regression parameters.

The temperature sensitivity of soil respiration (Q₁₀) represents the increase factor in rate of soil respiration when soil temperature is increased by 10 °C (Davidson, Janssens, & Luo, 2006; Yan et al., 2010). The Q₁₀ value was calculated by Equation 3, and ‘b’ was attained by Equation 1 when soil moisture had no significant influence on soil respiration; however, if soil moisture had a significant influence on soil respiration, the Q₁₀ value was calculated by Equation 4, and “b” was attained by Equation 2.

(3)

(4)

2.7 Statistical analysis

Data were analyzed on the basis of a randomized block design with community type and degradation type as factors. Repeated-measures ANOVA was used to analyze effects of year, community type, and degradation type for AGB, litter mass (LM), BNPP, and annual means of soil MBC, soil respiration components, T, and M. In addition, a general lineal model was used to determine effects of community type and vegetation degradation on measured variables in each year. Repeated-measures ANOVA was used to examine effects of vegetation degradation on soil respiration components for each community type in each year. Stepwise multiple regression analysis and curve estimation were performed to determine dependence of soil respiration components on soil temperature and soil moisture over the two experimental years. Stepwise multiple regression analysis was used to assess biotic and soil chemical control on soil respiration components. Pearson's correlation analysis was used to detect relationships between plant biomass and soil characteristics. Differences between means were detected with a least significant difference test. All data analyses were performed using SPSS 17.0 for Windows (USA), and statistical significance was defined as p = .05.

3.1 Vegetation and soil properties

AGB, BNPP, LM, and soil MBC significantly decreased as vegetation degraded, especially in C. virgata communities (Tables 2 and 3). SOC, soil TN, and TP concentrations under the moderate degradation (MD) type were reduced by 3.9%, 4.3%, and 3.6%, respectively, compared with those on the LD type for L. chinensis communities, and 3.8%, 4.6%, and 3.9% for C. virgata communities. For the SD type, SOC, soil TN, and TP concentrations were reduced by 14.1%, 10.9% and 10.7% compared with those on the LD type for L. chinensis communities, and 15.4%, 12.3%, and 11.9% for C. virgata communities (Figure 3a–c). Soil pH and soil bulk density significantly increased as vegetation degraded for both communities (Figure 2d and e). Compared with C. virgata communities, L. chinensis communities had higher values for AGB, BNPP, LM, soil MBC, and nutrients but had lower soil pH (Figure 3a–d). Across all degradation types, AGB, BNPP (0–10 cm), and LM were positively correlated with SOC, soil TN, and TP concentrations but negatively correlated with soil bulk density and pH at 0–10 cm of soil depth for both community types (Table 4).

3.2 Soil temperature, moisture, and respiration

Within growing season, soil temperatures under different degradation types in both communities had a consistent trend of SD > MD > LD (Figure 3a–d). Annual mean soil temperature in C. virgata communities, compared with L. chinensis communities, increased more significantly as vegetation degraded (Tables 2 and 3). Soil moisture had evident annual and seasonal changes (Tables 2 and 3, Figure 4e–h). Total soil moisture decreased as vegetation degraded in both communities, especially in drought seasons. Root respiration (SR_A) as well as microbial respiration (SR_H) under different degradation types had clear seasonal patterns for each community, which totally decreased as vegetation degraded in the whole-growing season (Figure 5). Annual mean SR_A and SR_H in 2012 were 23–44% and 25–40% higher than those of 2013 for all degradation types in both communities (Tables 2 and 3). Over the two growing seasons, annual mean SR_A and SR_H for all three degradation types were 88–94% and 57–73% higher in L. chinensis than in C. virgata communities. Depending on community type and year, vegetation degradation induced a significant decrease in annual mean SR_A. Independent of year and community type, annual mean SR_H significantly decreased as vegetation degraded (Tables 2 and 3).

3.3 Water and temperature controls on seasonal soil respiration

At interseasonal scales, stepwise multiple regression analysis indicated that both soil temperature and moisture affected SR_A of L. chinensis communities; however, SR_A of C. virgata communities only was regulated by soil temperature (Table 5). Both soil temperature and moisture had important effects on SR_H for two communities. For two respiration components, only Q₁₀ of SR_A had a significant difference between the two community types (Figure 6).

3.4 Abiotic and soil chemical controls on soil respiration components

AGB was the only biotic control factor of SR_A and SR_H for two communities, which accounted for the 89% and 93% of the variation in SR_A and 90% and 91% of the variation in SR_H for L. chinensis and C. virgata communities, respectively (Figure 7a and 7b). The SOC was the only soil chemical control factor of SR_A and SR_H for the two communities, which accounted for 66% and 76% of the variation in SR_A and 74% and 75% of the variation in SR_H for L. chinensis and C. virgata communities (Figure 7c and 7d).

4.1 The influences of vegetation degradation on soil respiration components

Rates of root and microbial respiration were significantly higher in L. chinensis than C. virgata communities under the same vegetation degradation gradient, which was attributed to better vegetation and soil conditions in L. chinensis communities compared with C. virgata communities (Wang et al., 2015), including (a) higher AGB, root production, and biomass (Yu et al., 2013); (b) higher soil substrate availability and soil microbial biomass (SOC and litter); and (c) higher soil water and nutrient content (Table 2), which directly or indirectly enhanced root and soil microbial activities (Yan et al., 2010).

Vegetation degradation induced by grazing disturbance has important effects on plant biomass and soil properties of grassland (Adewopo et al., 2015; Gong et al., 2014). In many studies, including the present one, AGB and root production significantly decreased as grazing pressure increased in a grassland ecosystem (Bai et al., 2015; Rey et al., 2011). Further, grazing induced vegetation cover decline and likely was responsible for changes in soil physicochemical properties. For example, declined vegetation cover could increase soil temperature and decrease soil moisture by enhancing surface radiation and evaporation (Song et al., 2013). In addition, declined vegetation cover could reduce soil substrate quality, substrate availability, and soil microbial biomass by litter input and decomposition (Ward et al., 2015). The change in plant biomass and soil properties could also influence root respiration and microbial respiration. For example, Yan et al. (2010) reported that enhanced AGB and BNPP increased root respiration and that soil MBC increased microbial respiration following addition of nitrogen and water. In another study, SOC concentration was positively correlated with soil respiration rate along a land-use sequence (Gong et al., 2014). In the current study, both root and microbial respiration in two meadow communities significantly decreased as vegetation degraded over a 2-year interval, consistent with a report on an alpine meadow ecosystem (Cao et al., 2004). However, there were no significant differences in soil respiration rates associated with ground cover in a subtropical steppe (Rey et al., 2011); therefore, perhaps effects of vegetation degradation on soil respiration are affected by grassland type. Further, on the basis of regression analysis, AGB and SOC had principal biotic and soil chemical controls of root and microbial respiration, which suggested that grazing disturbances decreased soil CO₂ release by reducing AGB and SOC concentration in this meadow ecosystem. Although soil CO₂ efflux is regulated by many environmental and biotic factors, we speculated that these two factors have more close and widespread associations with other factors and are more representative of root and microbial activities in this meadow ecosystem. AGB and SOC explained 66–93% of the variation in root and microbial respiration for the two meadow communities; therefore, they could be useful indicators for assessing soil respiration in this meadow steppe ecosystem.

Averaged across two growing seasons, root respiration decreased more significantly as vegetation degraded for C. virgata communities compared with L. chinensis communities, although L. chinensis communities had better vegetation and soil conditions. With similar responses to root respiration, AGB and SOC also had more significant decreases as vegetation degraded in C. virgata communities than in L. chinensis communities. There were significant interactions between community type and degradation type on AGB and SOC, consistent with larger changes in AGB and SOC likely causing a more sensitive root respiration response when meadow communities were disturbed by grazing. In contrast to root respiration, microbial respiration in both communities had a similar response to vegetation degradation. Perhaps grazing disturbance had equal effects on SOC decomposition of different communities in this meadow ecosystem, despite differences in properties and changes of vegetation and soil among communities.

4.2 Temperature, moisture control, and Q₁₀ of soil respiration components

Climate can drive great interyear changes in soil respiration by affecting soil temperature and moisture (Suseela, Conant, Wallenstein, & Dukes, 2012; Yan et al., 2010). In this study, annual mean SR_A and SR_H were 23–44% and 25–40% higher in 2012 than in 2013 for all degradation types in both communities, due to higher precipitation in 2012. Moreover, differences in rainfall can have variable effects on soil respiration among meadow communities. In this study, changes in annual rainfall influenced root and microbial respiration more intensively in C. virgata communities than in L. chinensis communities, consistent with a more stable CO₂ efflux for the latter communities under changed rainfall. In addition, rainfall change may alter effects of grazing disturbance on soil respiration. For example, in a rainfall-rich year, grazing resulted in significantly lower soil respiration rate than did grazing exclusion in a temperate typical steppe, although the opposite occurred in a relatively dry year (Gong et al., 2014). In this study, root respiration was more significantly reduced by grazing under relatively dry conditions (Table 2). It was reported that land degradation could alter the temporal pattern of soil respiration (Rey et al., 2011). However, in the current research, microbial respiration had similar seasonal changes across three degradation types for each community, which suggested climate-independent control of microbial respiration on a seasonal scale (Scott-Denton, Rosenstie, & Monson, 2006). Further, in this study, soil temperature and moisture jointly influenced seasonal microbial respiration of two communities, and seasonal root respiration of L. chinensis communities. However, seasonal root respiration of C. virgata communities was only regulated by soil temperature, probably because root phenology and activity of C. virgata as annual plants were not closely related to rainfall during the growing season. In soils, temperature and moisture covary at spatial and temporal scales and soil temperature can decrease as moisture increases during the growing season, or vice versa (Liu & Luo, 2011; Sierra, Trumbore, Davidson, Vicca, & Janssens, 2015). Strong covariation between soil temperature and moisture at temporal scale may have a profound influence for predicting soil respiration responses under changing temperatures (Davidson, Belk, & Boone, 1998; Sierra et al., 2015). Yan et al. (2010) reported no significant correlation between soil respiration components and soil temperature when calculations used soil temperature as the single dependent variable; however, both root and microbial respiration had significant correlations with soil temperature combined with moisture. Similarly, in this study, compared with just soil temperature, the combination of soil temperature and moisture accounted for significantly more of the seasonal variation in microbial respiration (Table 4), which also suggests that soil temperature and moisture affect microbial respiration (Davidson et al., 1998).

In this study, vegetation degradation did not significantly change temperature sensitivity (Q₁₀) of microbial respiration for either community; therefore, differently degraded vegetation had equal effects on SOC decomposition in the context of foreseeable global warming. However, other studies reported different results. For example, Paz-Ferreiro, Medina-Roldán, Ostle, McNamara, and Bardgett (2012) suggested that land degradation induced by grazing intensified the positive feedback between SOC decomposition and climate in grasslands of northern England. In addition, Chuckran and Frank (2013) reported that grazing decreased Q₁₀ of SOC decomposition in grasslands of Northern United States. Many factors can cause differences in Q₁₀ response to land degradation induced by grazing, including soil texture, grazing history, land degradation status, and method of measuring CO₂ efflux (Chuckran & Frank, 2013). Consequently, additional studies are needed to separate these effects. The Q₁₀ of root respiration was significantly different between the two community types, suggesting that warming could induce a larger increase of root respiration in C. virgata compared with L. chinensis communities. However, it remains unclear whether and how future warming will influence root carbon input into soil in the two communities, as changes in root carbon assimilation in response to warming are not well understood for these two communities.

4.3 The potential mechanism of land degradation and implication on grassland management

Livestock trampling increases as grazing duration and intensity increase, which can compact soil and increase soil bulk density, as what occurred in the current study. Increased soil bulk density is an important indicator of land degradation in grasslands, and it can alter other soil traits, including rainfall infiltration and water retention and thereby affect plant establishment and growth (Li, Zhou, et al., 2014; Orgill et al., 2016; Tóth et al., 2016). Exclusion of grazing can overcome effects of livestock trampling and restore vegetation, in particular restore plant roots, thus decreasing soil bulk density (Table 3; Li, Zhou, et al., 2014). Frequently, vegetation degradation induced by grazing decreases land cover, which increases soil evaporation by enhancing surface temperatures and windspeed and probably promotes salt-rich ground-water movement from deeper soil layer to soil surface, increasing surface soil pH in this saline-alkaline meadow (Li, Zhou, et al., 2014; Wang & Ripley, 1997). Under grazing, animal browsing transfers soil nutrients into livestock systems and decreases nutrient return by reducing litter input, which induces soil nutrient loss (Table 4; Pei, Fu, & Wan, 2008; Li, Zhou, et al., 2014). In that regard, soil TN and TP concentrations decreased as vegetation degraded. Generally, soil carbon storage is determined by the balance between biomass carbon input and SOC decomposition; the latter is substantially affected by root and microbial respiration. Microbial activity directly consumes soil organic matter, so increasing microbial respiration can enhance SOC decomposition (Conant et al., 2008) and decrease SOC storage. Root respiration reduces root biomass accumulation and carbon input from root into soil, which alters substrate availability for soil microorganisms, and likely reduces the priming effect of root carbon input on microbial activity of soil (Xiao, Guenet, Zhou, Su, & Janssens, 2015), thereby affecting microbial respiration and SOC decomposition. In this study, for both communities, SOC storage and root and soil microbial respiration synergistically decreased as vegetation degraded; consequently, decreased SOC storage was primarily attributed to decreased biomass carbon input (Table 2). Likewise, the more sensitive response of biomass input may drive more significant changes in SOC as vegetation degraded for C. virgata compared with L. chinensis communities. However, vegetation degradation also likely induces other mechanisms promoting soil degradation, for example, soil erosion (Lal, 2003). In the Songnen Plain, soil erosion may induce significant loss of soil nutrients and cause subsoil with higher salt content and lower nutrient concentrations to be exposed to the surface (Li, Yu, Li, Zhou, & Chen, 2014; Yu et al., 2013). Moreover, soil erosion may change the microbial decomposition of SOC by altering soil layer distribution and destroying soil aggregates (Häring, Fischer, Cadisch, & Stahr, 2013; Mchunu & Chaplot, 2012). As above, soil erosion may greatly confound assessment of carbon balance and land degradation in this grassland ecosystem (Lal, 2003; Li, Yu, et al., 2014). Therefore, further research is necessary for distinguishing between soil erosion and bio-driven processes in soil carbon loss and land degradation.

Overgrazing has induced intense vegetation degradation in Songnen grassland (Li, Zhou, et al., 2014). The grassland degradation further restricted the local economy and development of society. However, it is difficult to sustainably manage grassland to avoid grassland degradation and concurrently sustain milk and meat production. On the basis of present results, maintaining vegetation cover and biomass input was important for soil nutrient retention and suppression of soil erosion. Long-term grazing induced very intense vegetation degradation at the SD site, which greatly suppressed nutrient return, and probably increased soil erosion risk, likely accompanied by severe soil degradation. Therefore, grazing exclusion would be necessary for restoring vegetation cover at an SD site. Vegetation and soil were highly vulnerable in this grassland ecosystem, even under moderate grazing pressure, as after only 4 years of moderate grazing (stocking rate of 2 sheep·ha⁻¹), vegetation and soil in the MD site were clearly more degraded than an area with LD status (Tables 1 and 2, Figure 3). Therefore, <2 sheep·ha⁻¹ or rotational grazing should be recommended in this meadow ecosystem, reducing vegetation loss and ensuring enough time for plant regrowth. However, a decreased stocking rate would likely restrict livestock production in this grassland. Alternatively, increasing forage production and increasing grazing tolerance of grassland are likely better approaches. With our results, the vegetation and soil of L. chinensis communities had greater productivity and grazing tolerance than had C. virgata communities. Overseeding with L. chinensis to replace C. virgata communities would be expected to improve forage production and soil carbon and nitrogen storage of this grassland, even if it will possibly induce increase in soil respiration. However, higher soil salt and less soil nutrient content are two key limiting factors for L. chinensis establishment in C. virgata patches. Soil desalination and salt-tolerant L. chinensis breeding, supplemented by soil fertilization, could eliminate the bottleneck of L. chinensis establishment in C. virgata patches. Legumes are usually high-quality forage and can increase N availability in soils by symbiotic N₂ fixation and N transfer, thereby enhancing forage production and forage nutritive value (Li et al., 2015; Mortenson, Schuman, Ingram, Nayigihugu, & Hess, 2005). Recently, many studies, including some in this grassland, have indicated that legumes can promote soil C and N sequestration by increasing biomass C and N input (Fornara & Tilman, 2008; Li, Yu, Li, & Zhou, 2016; Wu, Liu, Tian, & Shi, 2017). Legumes have positive effects on forage production, and soil C and N sequestration, thus improving grassland sustainably (Li et al., 2016). Although legume abundance in this grassland is very low (Table 1), legume introduction would be expected to improve forage and livestock production and increase grassland sustainability. In addition, the Songnen grassland is located at agro-pastoral transitional zone; the forage and animal production can be assured by planting feed crops and developing artificial grassland in low-fertility cropland, which are being greatly supported by policy and finance from China's Government. Also crop straw resource is abundant in this region. The crop straws, such as maize straws, can be beneficial supplement to forage if they are utilized enough.