Potential productivity of the Miscanthus energy crop in the Loess Plateau of China under climate change

The Loess Plateau extends from central to northwestern China (34°–45° 5'N, 101°–114° 33'E). It belongs to the warm or temperate continental monsoon climate, which is dominated by the East Asian Monsoon (Chen et al 2007, Fu et al 2009, Wang et al 2011). The average annual temperature ranges from 6 to 10 ° C from northwest to southeast and the annual precipitation is largely between 300 and 600 mm. In particular, annual precipitation is unevenly distributed in the Loess Plateau, with the majority of precipitation occurring in July and August. The region spans from arid, semiarid to semihumid zones and therefore is considered to be a semiarid to semihumid transitional zone which is very sensitive to climate change.

Regional climate scenarios were generated by the regional climate change model PRECIS (Providing Regional Climates for Impacts Studies) (Jones et al 2004, Xu 2004, Zhang et al 2006). PRECIS took the output of the global model, HadAM3H, at its lateral boundaries using a high resolution (50 km × 50 km grid). There were 262 grid cells covering 62.4 Mha of the Loess Plateau. PRECIS simulated daily values of temperature, radiation, and precipitation in two periods, the baseline from 1961 to 1990 and the future from 2011 to 2099. Its performance in the simulation of China's climate was validated (Xu et al 2007). The scenarios of CO₂ concentration followed the B2, A1B, and A2 emission scenarios, which presented low–medium–high emission based on the IPCC SRES storylines (Nakicenovic and Swart 2000). The CO₂ concentration scenarios in PRECIS were derived from SRES emission scenarios which used a simple carbon cycle model and did not consider a potential acceleration in increasing CO₂ concentrations due to climate–carbon cycle feedbacks.

Because different climate models simulated different response patterns in temperature and precipitation, it was important to compare the behavior of PRECIS with those from other climate models in order to reflect this source of potential uncertainty. A total of 17 climate models were available through the Program for Climate Model Diagnosis and Intercomparison (PCMDI) for the IPCC Fourth Assessment Report (Meehl et al 2007). Climate scenarios from PRECIS produced warming similar to the average from all of the 17 models for China but produced wetter conditions than the average from these models for China (Xiong et al 2008). A2 PRECIS predicted that precipitation increased 5% for the 2020s relative to the baseline, whereas the average of all models was 0.5%. For the 2050s, PRECIS predicted that precipitation increased 10%, while the all-model average was 3.6%. For the 2080s, it predicted that precipitation increased 15%, while the all-model average was about 10%.

A yield model was previously developed for estimating the yield potential of Miscanthus energy crops in the Loess Plateau based on the yield of Miscanthus lutarioriparius measured in an experimental field located in Qingyang, Gansu Province in China in 2010 (Liu et al 2012b). This yield model was derived from the radiation model (Monteith 1977), especially with reference to its application in the field study of M. × giganteus (Beale and Long 1995). In the study of M. lutarioriparius, the model incorporated the variation of growing season length described by a nonlinear relation of annual accumulated temperature over 10 ° C. Water limitation reflected by precipitation was also taken into consideration, given that the Loess Plateau is largely an arid and semiarid area. According to the field data, we estimated the yield potential in Qingyang in 2010 to be 22.5 t ha⁻¹ yr⁻¹ after a two-year field experiment (Liu et al 2012b, Yan et al 2012). Using this model, we estimated the productivity of the energy crop each year in the baseline period (1961–1990) and the future period (2011–2099) at each grid cell using the annual incident radiation, annual accumulated temperature over 10 ° C (AT₁₀), and annual precipitation at each grid cell in the Loess Plateau in the two periods.

Note that the yield was set to zero in the grid wherever annual precipitation was less than 250 mm or AT₁₀ was less than 2000 degree days (Liu et al 2012b). Because the radiation data provided in the three climatic scenarios were not validated, we used the observed average radiation from 1961 to 1990 in each grid cell in the Loess Plateau to scale the radiation in the baseline. We also used the observed radiation in Qingyang from 2011 to 2012 to scale the radiation for the future period in order to be consistent with the radiation in Qingyang in 2010. Because considering land-use change over the 90-year period was not possible, this study did not include land quality in the yield model. As a result, this study focused on the impact of climate change on Miscanthus yield and the trend of potential yield rather than an estimate of the total potential yield in the Loess Plateau.

To evaluate the validity of the yield model that was developed based on biomass yield in the 2010 growing season, we tested the consistency between the measured and model-predicted yield for the 2011 and 2012 growing seasons. At the end of the growing seasons, plant height and tiller diameter of M. lutarioriparius planted in the Qingyang experimental field was measured. The biomass of these individuals was determined based on the relationship established from a linear regression analysis between biomass and plant volume obtained for 112 individuals in the 2010 growing season (Yan et al 2012). The plant volume was calculated from the plant height and diameter measured at the end of the growing season as a cone shape. The yields of 2011 and 2012 were predicted by the model using the climatic data of the year, including annual incident solar radiation, annual accumulated temperature, and annual precipitation. The same 112 individuals were compared for the three consecutive years using a t-test.

Data for the annual incident solar radiation, annual accumulated temperature, and annual precipitation in 2010 and the radiation in 2011 and 2012 in Qingyang were obtained from daily data from the Xifeng Metrological Station adjacent to the Qingyang experimental site (Gansu, China). The radiation from 1961 to 1990 in the Loess Plateau was obtained from the China Meteorological Data Sharing Service System (http://cdc.cma.gov.cn/). Daily simulated radiation, temperature, and precipitation across the Loess Plateau in the baseline period from 1961 to 1990 and the future period from 2011 to 2099 were taken directly from RegCM3 output from Chinese Academy of Agricultural Sciences and used as the input necessary to drive the yield potential model above. All climate scenario datasets were compiled at a resolution of a 50 km × 50 km grid. The data were converted from geographic coordinates to projected coordinates in order to align them spatially and then assembled according to their geographical midpoints.

Given all the data described above, the annual yield potential at each grid cell in the Loess Plateau was calculated for the two periods using the yield model. Based on the simulated 90-year (from 2011 to 2099) yield potential dynamics, we calculated the trend of average yield change in the Loess Plateau and the trend of yield change at each grid cell using the linear regression method. Linear regression analysis was used to identify trends and the relative contributions of climatic variables in the simulated yield time series (Xiong et al 2012). The relative increasing percentage rate was also calculated, which was the increasing trend divided by the average yield. The trends of the change of annual precipitation, annual radiation, and annual AT₁₀ were calculated at each grid cell in the future period using linear regression. All statistical analyses were conducted using R 2.15.1 for Windows (R Development Core Team 2011). We then used marginal analysis to identify the degree of the impacts of three climatic variables, annual precipitation, annual radiation, and annual AT₁₀, on yield using the partial derivative of the yield. The maps of yield, the trend of yield change, the relative rate of yield change, the effect of three climatic variables on yield, and the trend of climate change were generated by interpolating between 262 grid cells in the Loess Plateau.

Under the climate change scenarios, our model predicted that Miscanthus yield in the Loess Plateau increased in the future. Comparing with the baseline, the average yield at all grid cells in the Loess Plateau would increase 10.3%, 9.1%, and 12.6% in A2, B2, and A1B scenarios, respectively (figure 2). The average yield potential of the baseline was 16.9 t ha⁻¹ yr⁻¹, very close to the value of 16.8 t ha⁻¹ yr⁻¹ estimated previously using the climate data of the past 30 years (Liu et al 2012b). This indicated that the climate parameters inferred from the climate change model and validated by the previous studies were reasonably close to the real situation in the past.

In addition to the increase in the average yield potential, the area available for growing Miscanthus energy crops would also increase in the Loess Plateau for the future period. In comparison with the baseline, the area of land potentially available increased by 10.1%, 7.3%, and 8.9% in A2, B2, and A1B scenarios, respectively (figure 2). Because the land available and average yield both contributed to energy crop production, the increase in the total yield of Miscanthus reached high percentages for the entire Loess Plateau, 21.6%, 17.0%, and 22.4% in A2, B2, and A1B scenarios, respectively.

This impressive level of yield increase was at least partly owing to trends of increase that continued without major setback throughout the future period (figure 3). In contrast, the previous studies reported that the production of Miscanthus in Europe would decrease from 2020 or 2050 under various scenarios due primarily to a reduction in precipitation (Hastings et al 2009, Tuck et al 2006). Thus, the Loess Plateau would be a desirable place for the large-scale production of energy crops from now into the future.

The marginal analyses showed that an increase in precipitation had the largest impact on yield, followed by an increase in temperature in the Loess Plateau. Changes in radiation had little effect on the yield. Specifically, the average contribution of annual precipitation to yield was 0.0375, i.e., 1 mm precipitation led to an additional 0.0375 tons of biomass per hectare per year. This was twice as much as the contribution of a temperature increase. The Loess Plateau mainly consists of arid and semiarid areas, such that water shortage has been the key limiting factor for the yield of food crops that rely exclusively on rainfeed. The relative increase was especially sharp in the western part of the Loess Plateau, which is relatively dry and cold at the present time.

The area available for growing Miscanthus would increase mainly because some parts change from unsuitable to suitable for energy crop production. The land was assumed to be unsuitable if annual precipitation was less than 250 mm or AT₁₀ was lower than 2000 degree days (Liu et al 2012b). With the climate change, the trend of increase in precipitation and AT₁₀ would change land use mainly from unsuitable to suitable. The combination of the two factors also led to such a change concentrating in the arid–semiarid transition zone which marginally supports energy crop production under the present climatic conditions.

The study of the impact of climate change on food crop production in this region also predicted a generally increased trend of precipitation and yield. Li et al (2011) used the WEPP model and reported a change of −17–25% for wheat yield and −2–39% for maize yield in the next 40 years using the projection of −2.6–17.4% change of precipitation. Another study reported a similar result, but the food yield would increase slightly in the next 40 years, along with a 23–37% predicted increase in annual precipitation (Zhang and Liu 2005). In comparison with food crops, the energy crop would have a more sensitive response to the increase of precipitation. This is likely because the biomass yield has a better linear relationship with water availability under rainfed conditions than grain yield. Moreover, the positive response to the temperature increase is probably also quite unique to the C4 energy crop. A comparison of the response of food and energy crops to climate change will be an interesting topic to study for a region such as the Loess Plateau where both crops are likely to coexist.

Under all scenarios, atmospheric CO₂ concentration will increase in the future. Because such an increase is likely to have rather little impact on the productivity of C4 plants (Leakey et al 2009, de Souza et al 2013), CO₂ concentration was not incorporated into the yield model. Nevertheless, there have been reports that the change in CO₂ concentration might play a role in water uses, such as evapotranspiration, which in turn indirectly affect photosynthesis and yield, especially under drought conditions (Leakey et al 2009, Hussain et al 2013). Future studies focused on the impact of growing Miscanthus energy crops on soil water content and even the regional hydrologic cycles are needed to paint a more complete picture of yield changes under climate change (Le et al 2011).

In a previous study using the recorded climate data of the past 30 years, it was estimated that an area of nearly 49 Mha could potentially support Miscanthus production in the Loess Plateau (Liu et al 2012b). More than 15 Mha of this was considered to be high-quality cropland, pasture or land suitable for afforestation, which should be dedicated for growing food crops to ensure food security. Under the scenarios of climate change, land that could potentially support Miscanthus production increased to about 57 Mha by the end of the century. This was due primarily to the change of land incapable of supporting Miscanthus production into land suitable for growing the energy crop as a result of increases in annual temperature and/or precipitation. Meanwhile, it was inevitable that land previously suitable for growing energy crops would change into high-quality cropland or pasture with the climate change, which would reduce the total area of marginal land that can actually be utilized for producing energy crops. This trend of land-use change, however, was not estimated in this study because more detailed land-use criteria for food and energy crop production had yet to be established.

In any event, land-use change in either direction would generate positive environmental effect in the Loess Plateau. Growing perennial energy crops on land with relatively poor natural conditions can lead to additional soil carbon accumulation and the improvement of soil structure and function (Lal 2004, Matamala et al 2008, McLauchlan et al 2006, Post and Kwon 2000). Restored soil fertility will further reduce soil erosion and improve land productivity. In the case that marginal land used for energy crop production is converted to cropland with the climate change, the land would benefit from growing perennial energy crops in terms of soil improvement. In essence, energy crop production for at least a period of time in the region suffering from serious erosion would function as means of ecological restoration. This, combined with the climate change, would create a larger area of high-quality cropland in the Loess Plateau, currently one of the most seriously eroded regions in the world.

Thus, the interchange between land for food and energy crop production may contribute to the development of a more sustainable agricultural system in regions where land had been overused for growing annual food crops and suffered from erosion and degradation. The rotation between food and perennial energy crops, especially with climate change taken into consideration, would give rise to a new agricultural system balancing food and energy crop production in the long run.

In addition to this overall strategy, this study provides a map of energy crop yield trends under scenarios of climate change. Across the region, it could form a basis for guiding long-term bioenergy development and ecological restoration in the Loess Plateau. Obviously, there is still much room for further analyses that would ensure more accurate and detailed guidance for economic and environmental development. There is a need to establish more detailed and grounded models based on an understanding of the response of Miscanthus production to climate change. In particular, the availability of additional sources of climate change scenarios at a finer resolution in the regional climate model will help in evaluating and improving the analyses.