See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/240797895

Estimation of pasture productivity in Mongolian grasslands: Field survey and model simulation

Article in Journal of Agricultural Meteorology · January 2010 DOI: 10.2480/agrmet.66.1.6

CITATIONS READS 11 336

3 authors:

Tserenpurev Bat-Oyun Masato Shinoda Institute of Meteorology, Hydrology and Environment Nagoya University

14 PUBLICATIONS 95 CITATIONS 132 PUBLICATIONS 1,859 CITATIONS

SEE PROFILE SEE PROFILE

Mitsuru Tsubo Tottori University

116 PUBLICATIONS 2,727 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Migration ecology and conservation of Mongolian wild ungulates View project

Sand fluxes and its vertical distribution in the southern : A sand storm case study for 2011 View project

All content following this page was uploaded by Tserenpurev Bat-Oyun on 06 January 2014.

The user has requested enhancement of the downloaded file. Full Paper J. Agric. Meteorol. (農業気象) 66 (1): 31-39, 2010

Estimation of pasture productivity in Mongolian grasslands: field survey and model simulation Tserenpurev BAT-OYUN†, Masato SHINODA, and Mitsuru TSUBO (Arid Land Research Center, Tottori University, Hamasaka, Tottori, 680–0001, Japan)

Abstract The Mongolian economy depends critically on products of range-fed livestock. Pasture is the major food source for livestock grazing, and its productivity is strongly affected by climatic variability. Direct measurement of pasture productivity is time-consuming and difficult, especially in remote areas of a large country like Mongolia with sparse spatial distribution of pasture monitoring. Therefore, model- ing is a valuable tool to simulate pasture productivity. In this study, we used a remote sensing-based production efficiency model, that is, the Carnegie Ames Stanford Approach (CASA) model, to estimate pasture productivity in three main vegetation zones of Mongolia; desert steppe, steppe and forest steppe. The present study aimed to explore climatic and grazing effects on grassland productivity in Mongolian grasslands during 2005-2007, using ground-based measurements and simulation model outputs. The ground measurements showed that grazing caused a significant decrease in measured aboveground phytomass and plant height. Simulation results demonstrated that the highest net primary productivity (NPP) of 83.2 gC/m2 and the lowest NPP of 12.6 gC/m2 over the growing season (April-September) occurred at Darkhan (steppe) and (desert steppe), respectively. Moreover, the comparison of temperature and water stresses on pasture productivity indicated that water stress was stronger down- regulator of NPP, verifying that drought is the major concern of pasture production. Based on the comparison between the measurements and simulation, the ratio of aboveground NPP to belowground NPP in the Mongolian perennial grasslands was estimated as 1:1.5. Key words: Aboveground phytomass, Drought, Grazing, Net primary productivity, Water stress.

Ni, 2003; Shinoda et al., 2007; Suzuki et al., 2007; 1. Introduction Zhang et al., 2005). Pastoral animal husbandry plays a key role in the Continuous monitoring and modeling of grasslands' Mongolia's economy, producing 40％ of gross domestic production can provide information on feed availability, product. Production and growth of livestock are greatly efficient management of livestock grazing, and natural dependent on the productivity of natural grasslands, hay preparation for winter; this information is very which cover 80％ of the country (Batima and Dagva- useful for herders and decision-makers. However, due dorj, 2000). Currently, about half the population is to the large area of Mongolia, it is difficult to monitor dependent on livestock production for their livelihood vegetation conditions widely, using field measurements (Johnson et al., 2006). The production of grassland at specific locations, which makes it difficult to gain is regulated by many factors, such as precipitation, a comprehensive understanding of how the grasslands temperature, solar radiation, soil nutrient availability, respond to factors such as climate, grazing, fire and and grassland utilization and management. Water stress management. Modeling is an essential approach to is the most limiting factor for dryland vegetation due to improve our understanding of the complex dynamics of the region's low precipitation and high evapotranspira- ecosystems such as grasslands. In Mongolia, progress tion (Miyazaki et al., 2004; Munkhtsetseg et al., 2007; in estimating and modeling the carbon cycle of grass- land ecosystems has been seriously limited due to the Received; April 13, 2009. scarcity of observational data for the parameterization Accepted; September 9, 2009. and validation of model and the substantial uncertain- †E-mail: [email protected]

31 J. Agric. Meteorol. (農業気象) 66 (1), 2010 ties in net primary productivity (NPP) estimations pasture productivity in Mongolia, using ground-based based on field measurements. measurements and the CASA simulation model based

Nakano et al. (2008) measured CO2 fluxes both on satellite data. inside and outside of a site of drought experiment that 2. Materials and Methods was conducted at Bayan Unjuul in Mongolia (Shinoda et al., 2009), using a closed-chamber technique. They 2.1 Study sites demonstrated that the reduction of gross primary Fig. 1 illustrates four study sites; Mandalgovi production per unit aboveground biomass (GPP/AGB) (45.77°N, 106.28°E, desert steppe), Bayan Unjuul was caused by a combination of high vapor pressure (47.04°N, 105.95°E, steppe), Darkhan (49.47°N, deficit and low soil moisture. Bolortsetseg (2006) 105.98°E, steppe), and (48.80°N, 103.55°E, analyzed the effects of climate change on AGB forest steppe) located in three vegetation zones of in Mongolia using the Century model (namely, a Mongolia. In general, during the growing season (April- model designed to simulate carbon, nutrient, and water September), precipitation decreases and temperature dynamics for different types of ecosystems including increases from the north to south, resulting in warmer grasslands). The sensitivity analysis indicated that AGB and more arid conditions in the south. The growing was more sensitive to changes in precipitation than season for the grasslands is very short and is limited those in temperature. by low temperature and precipitation. In particular, A number of models have been developed to simulate moisture availability is generally considered the most productivity of different ecosystems; among these important determinant for vegetation growth and the we selected the Carnegie Ames Stanford Approach large seasonal variations in precipitation are clearly (CASA) model, because it does not include complex reflected in plant growth (Gunin et al., 1999). ecophysiological parameters. The NPP component of Average (1995-2007) air temperature at the Institute the CASA model is based on the concept of radiation of Meteorology and Hydrology (IMH) station was use efficiency (RUE). In origin, Monteith (1972) 14.4℃ at Mandalgovi, 14.3℃ at Darkhan, 13.8℃ at explored that primary production was linearly related Bayan Unjuul and 11.8℃ at Bulgan during the growing to the amount of radiation received by plant stand. season. Average precipitation during the growing Later, this approach was theoretically and experimen- season was low at Mandalgovi (124 mm) and Bayan tally strengthened (Monteith, 1977); as a result, it has Unjuul (135 mm) and high at Darkhan (268 mm) and become the most commonly used method of analyzing Bulgan (291 mm). It should be noted that the study and modeling plant growth. It has been found that a years (2005-2007) were among the driest years during remote-sensing-based vegetation index is an indicator 1995-2007 for the study sites except for Darkhan. of absorbed photosynthetically active radiation (APAR) Plant composition varies among the different by green vegetation (Sellers, 1987; Sellers et al., 1992; ecosystems. The Bulgan and Darkhan sites are co- Goward and Huemmrich, 1992). In the CASA model, dominated by Stipa krylovii, Agropyron cristatum, the fraction of incoming photosynthetically active radia- Cleistogenes squarrosa, Leymus chinensis, and Carex tion (PAR) intercepted by green vegetation (FPAR) is spp., while Mandalgovi is co-dominated by Stipa estimated from normalized difference vegetation index (NDVI) data (Potter et al., 1993). Several studies have evaluated the performance of the CASA model for various regions. For instance, the CASA model successfully simulated the spatial and temporal distribution of NPP in northern China using Moderate Resolution Imaging Spectroradiometer (MODIS) data (Yuan et al., 2006), interannual variations of cropland NPP in the (Lobell et al., 2002), and maize production across China (Tao et al., 2005). However, there have been very few attempts to apply this kind of model to Mongolia. Given this background, Fig. 1. Vegetation zone map of Mongolia. Study site we aim to investigate climatic and grazing effects on locations are shown as black dots.

32 T. Bat-Oyun et al. : Estimation of pasture productivity in Mongolian grasslands krylovii, Cleistogenes squarrosa, Allium polyrrhizum, hereinafter called height. and Artemisia frigida (Bolortsetseg et al., 2002). Sixteen-day composite MODIS NDVI images with The Bayan Unjuul site is co-dominated by perennial the highest resolution (250 m×250 m) were obtained grasses such as Stipa krylovii, Agropyron cristatum, from the Earth Observation System data gateway and Cleistogenes squarrosa, by forbs such as Artemisia (https://wist.echo.nasa.gov/api/). The products were adamsii and Chenopodium aristatum, and by small corrected atmospherically, considering ozone absorp- shrubs (Caragana spp.) (Shinoda et al., 2009). tion and molecular scattering. 2.2 Data 2.3 Model description Data used for this study were provided by IMH The CASA production-efficiency model was used of Mongolia. The data included climate (temperature, to simulate NPP (gC/m2/month), which downregulates precipitation, and solar radiation), soil moisture, photosynthetic efficiency in response to adverse soil texture, and aboveground phytomass (AGP) temperature and dry soil conditions (Potter et al., 1993; in no grazing plots (fenced exclosures to prevent Field, 1995). The model is based on the RUE model livestock grazing) and grazing plots (unfenced areas) described by Monteith (1972, 1977). NPP is estimated for Darkhan, Bulgan, and Mandalgovi during the from total PAR (MJ/m2/month), FPAR, and RUE as growing season of 2005-2007. No data were available expressed by the following equation: for Bulgan in the no grazing plot because of a missing NPP = PAR $FPAR $ RUE (1) observation. For Bayan Unjuul, we have measured the above variables, plus vegetation cover and plant Here, FPAR is calculated as a linear function of the height in no grazing and grazing plots since June simple ratio (SR). SR is given by: 2004. The experimental layout was a fully randomized SR x, t =1 + NDVI x, t / 1 - NDVI x, t design, consisting of two treatments (no grazing and ^ h 8 ^ hB 8 ^ hB grazing plots) with four replicates (quadrats). The (2) IMH measured animal-available AGP (located above where x represents the grid cell and t represents the the 1-cm height of grasses from the ground surface) month. FPAR is calculated as: in four 1-m2 plots at each of the three study sites FPAR x, t = min SR x, t / SRmax - SRmin (Darkhan, Bulgan, and Mandalgovi) at 10-day intervals ^ h # ^ h 6 @ -SRmin / SRmax - SRmin , 0. 95 for no grazing plots and at monthly intervals for 6 @ - grazing plots (except for the Bulgan no grazing plot). (3)

Unlike at the other stations, the clipping of AGP for where SRmax approximates the SR value at which all Bayan Unjuul was conducted at the ground level. The incident solar radiation is intercepted and it corrects leaf area index in the height of 0-1 cm accounts for the effects of canopy architecture and residual cloud approximately 10％ of the total for the forest steppe and contamination. This value is set to 5.13 for grasslands. steppe sites (Nachinshonhor, 2001). This suggests that SRmin represents the SR value for unvegetated land areas the animal-available AGP is a slightly underestimate of and is set to 1.08 for all grid cells. A cap of 0.95 was the total AGP above the ground level. The procedure imposed on FPAR in order to reflect a finite upper included clipping, drying and weighing the dried limit to leaf area. Thus, if FPAR is greater than 0.95, AGP. AGP was expressed as dry matter weight per FPAR is set to 0.95 in Eq. (3). unit area (g/m2). To compare the measured AGP with Radiation use efficiency is in turn calculated as the simulated results, phytomass values were converted product of the maximum RUE (RUEmax) and three into C equivalents (C values) using the ratio of 0.45 stress factors: (for grass and foliage components) as the mass of C RUE x, t = RUEmax $ Tf1 x, t per gram dry mass (Raich et al., 1991). For Bayan ^ h ^ h (4) $Tf2 x,,t $ Wf x t Unjuul, vegetation cover (ranging from 0 to 100％) ^h ^ h 2 was visually evaluated in four 4-m plots each for where Tf1 and Tf2 account for the effect of temperature no grazing and grazing plots. In the same plots, the stress and Wf accounts for effect of water stress. Tf1 height of the tallest individual of every plant species represents a physiological reduction of RUE when present was measured. The values of all species were temperatures are higher or lower than an optimum averaged for no grazing and grazing plots separately, temperature Topt, which is defined as the air temperature

33 J. Agric. Meteorol. (農業気象) 66 (1), 2010 in the month when the NDVI reaches its maximum were obtained using bilinear algorithm. NDVI images during the growing period. Tf1 ranges from 0.8 at 0℃ observed in grazing area were used to estimate NPP and 40℃ to 1.0 at 20℃ and it is given by: of grazing plots.

2 2.5 Simulation scenarios Tf1 x =0.. 8 + 0 02Topt x - 0.0005T x ^ h ^ h opt ^ h In order to reveal the sensitivity of simulated NPP (5) under various climatic conditions, we performed

Tf2 reflects the concept that the radiation use ef- simulations by changing stress conditions in Eq. (4) at ficiency should be depressed when plants are growing four stress levels; no stress (control excluding all the at temperatures displaced from their optimum. The stresses), temperature stress (water stress excluded), function has an asymmetric bell shape that falls off water stress (temperature stress excluded), and both more quickly at high than at low temperatures. Tf2 the temperature and water stresses (actual conditions ranges from 0 to 1 and it is calculated as: including all the stresses). Next, we quantitatively 1 estimated the relative contribution of each stress on Tf2 x, t = NPP for the four study sites. ^ h 1+ exp 0.2Topt x - 10 - T x, t $ _ ^ h ^ hi. $ 1 3. Results 1+exp 0.3 -Topt x -10 + T x, t $ _ ^ h ^ hi. 3.1 Climatic and grazing effects (6) 3.1.1 Effects of precipitation and grazing on

Calculation of Wf is based on a comparison between aboveground phytomass moisture supply (precipitation and stored soil water) In Mongolia, precipitation varies greatly from region and moisture demand. The water stress factor is to region. Total precipitation for the four months from calculated as: April to July (average of 2005-2007) was greater at Bulgan (181 mm) and Darkhan (170 mm) than at Bayan Wf x,.t =0 5 + 0., 5 $ EET x t /,PET x t ^ h ^ h ^ h Unjuul (58 mm) and Mandalgovi (56 mm). Similarly, (7) AGP was larger for Darkhan than for Bayan Unjuul where EET is estimated evapotranspiration and and Mandalgovi in both the no grazing and grazing PET is the potential evapotranspiration described by plots (Fig. 2). The mean AGP in July (the time of rapid

Thornthwaite (1948). Wf varies from 0.5 for very arid growth) was greater in the no grazing plots than in ecosystems to 1.0 for very wet ecosystems. Nemoto et the grazing plots at all the three sites. The differences al. (2003) pointed out that the original soil moisture between no grazing and grazing plots were significant submodel of the CASA model tends to produce a very (P＜0.05) for Darkhan and Bayan Unjuul, whereas it low variation in the relative drying rate (RDR which represents evaporation efficiency), resulting in very small and nearly constant evaporation at the Ordos site in Inner Mongolia, China. To solve this problem, they modified the equation of RDR so that it realistically produced a wide range of evaporation variation at the study site. Thus, we also used the submodel modified by Nemoto et al. (2003). The maximum RUE in CASA is set uniformly at 0.389 gC MJ–1 PAR (Potter et al., 1993); this value has been verified globally, comparing predicted annual NPP with more than 1900 field estimates of NPP (Potter et al., 2003). Fig. 2. Average aboveground phytomass in July 2.4 NDVI image processing from 2005 to 2007. Error bars represent standard In this analysis, we reprojected MODIS NDVI deviations. The asterisk denotes a significant dif- images from original Sinusoidal (SIN) projection to ference (P＜0.05) between aboveground phytomass more standard geographic map projection. We averaged values of no grazing and grazing plots. No data the pixel values of the four nearest neighbors that was available for the Bulgan no grazing plot.

34 T. Bat-Oyun et al. : Estimation of pasture productivity in Mongolian grasslands was not significant (P＞0.05) for Mandalgovi. respectively). In 2005, AGP, vegetation cover, and 3.1.2 Comparison of aboveground phytomass, veg- height were significantly larger in no grazing plots etation cover, and height between no grazing than in grazing plots (Table 1). However, in 2007, and grazing plots for Bayan Unjuul all the values were not significantly different between Based on our detailed measurements for no grazing the two types of plots. In 2006, AGP in no grazing and grazing plots, we examined the effects of grazing plots was significantly larger than that in grazing on AGP, vegetation cover, and plant height at Bayan plots, whereas vegetation cover and height were not Unjuul. AGP and height values averaged for Julies of significantly different. 2005-2007 were significantly lower (P＜0.05) in grazing 3.2 Model simulation plots than in no grazing plots. Nevertheless, vegetation 3.2.1 NPP in various vegetation zones cover was similar between the two types of plots (not Data from the four sites were analyzed to evaluate significantly different; P＞0.05). the effectiveness of model estimation of NPP for April to July precipitation was the highest in 2005 the different vegetation zones. The spatial pattern (65.3 mm, the wettest year among the three years) of simulated NPP (Fig. 3) was similar to that of and low in 2006 and 2007 (55.2 and 53.7 mm, measured AGP (Fig. 2), with the Darkhan (steppe) and

Table 1. Aboveground phytomass (AGP), vegetation cover (Cover), and plant height (Height) in no grazing (NG) and grazing (G) plots at Bayan Unjuul (steppe) in July of 2005-2007. Different letters (a, b) refers significant difference (P＜0.05) between AGP, Cover and Height of no grazing and grazing plots for each year. AGP (gC/m2) Cover (％) Height (cm)* Date NG G NG G NG G 21 July 2005 25.5a 9.8b 31a 20b 22a 7b 22 July 2006 15.0a 6.1b 13a 15a 14a 10a 28 July 2007 7.3a 8.2a 11a 10a 10a 5a *Height refers to the average of the heights of the tallest individuals of every spe- cies for no grazing and grazing plots

Fig. 3. Simulated cumulative monthly NPP (average of 2005-2007) from four separate runs: -no stress, -temperature stress, -water stress, -temperature and water stress of CASA over the four sites. Percentage to the right of each line shows the relative decrease over growing season compared to the control. The asterisk denotes actual condition of grassland productivity.

35 J. Agric. Meteorol. (農業気象) 66 (1), 2010

Bulgan (forest steppe) sites most productive and Bayan NPP included both the aboveground and belowground Unjuul (steppe) and Mandalgovi (desert steppe) less productivity (Fig. 4). We concluded that from the slope productive. The high values of simulated cumulative of linear regression between the measured AGP and NPP during the growing season (actual condition simulated NPP, approximately 40％ of total NPP was considering all the stress factors) were 83.2 and 82.6 stored in AGP and approximately 60％ in belowground gC/m2 for Darkhan (Fig. 3c) and Bulgan (Fig. 3d), phytomass (BGP). respectively, while low values were 33.2 and 12.6 gC/m2 for Bayan Unjuul (Fig. 3b) and Mandalgovi (Fig. 3a), respectively. 3.2.2 Effects of temperature and water stresses on NPP Responses of NPP to various stress levels were tested over the different vegetation zones, using the model output from 2005 to 2007. As expected from the latitudinal climatic gradient, Tf2 and Wf were relatively low for Bayan Unjuul and Mandalgovi, compared with Darkhan and Bulgan, due to insufficient water and unfavorable temperature conditions (Table 2). It can be seen that Tf1 was less variable, maintaining the highest values (no impact on NPP). Cumulative NPP over the growing season was reduced by 24-29％ due to the temperature stress and by 42-44％ due to the water stress for Bayan Unjuul and Mandalgovi, Fig. 4. Measured aboveground phytomass (AGP) in whereas for Bulgan and Darkhan, temperature and August and simulated cumulative NPP from April to water stresses decreased NPP by 15-18％ and 19-22％ August at four sites from 2005 to 2007. The gray of the control, respectively. These results demonstrate shaded area under regression line indicates AGP and that at all the sites, water was the primary factor that the white area between the 1:1 line and the regression influenced NPP (Fig. 3). line indicates belowground phytomass (BGP). 3.3 Comparison between simulated and measured 4. Discussion and conclusions results We compared the simulated results with the field- We revealed that both the grazing and precipitation measured AGP at the four sites. Since the NDVI pixel play important roles in determining pasture productiv- size (250 m×250 m) is much larger than the quadrat ity. In general, our results are consistent with the areas used for the field measurements (usually 1 m2), findings of previous studies on factors that affect plant we based simulated NPP on an NDVI pixel located growth; Fernandez-Gimenez and Allen-Diaz (1999) on in the grazing area and compared it with AGP of the biomass, Kondoh and Kaihotsu (2003) and Suzuki et grazing plots. Simulated cumulative NPP from April al. (2007) on NDVI, Miyazaki et al. (2004) on leaf to August was always larger than AGP in August area index, and Chen et al. (2007) on grazing. All (the time of peak phytomass), because the simulated these studies also reported that interannual changes

Table 2. Temperature stresses (Tf1, Tf2) and water stress (Wf) at four sites from 2005 to 2007 (averaged over the growing season). Temperature stresses Water stress Stress factors Tf1 Tf2 Tf1 Tf2 Tf1 Tf2 Wf Year 2005 2006 2007 2005 2006 2007 Mandalgovi (desert steppe) 1.00 0.76 1.00 0.73 1.00 0.79 0.56 0.58 0.54 Bayan Unjuul (steppe) 1.00 0.77 1.00 0.77 1.00 0.80 0.58 0.58 0.59 Darkhan (steppe) 1.00 0.81 1.00 0.79 1.00 0.85 0.80 0.83 0.75 Bulgan (forest steppe) 0.99 0.82 0.99 0.80 1.00 0.71 0.80 0.85 0.78

36 T. Bat-Oyun et al. : Estimation of pasture productivity in Mongolian grasslands in plant growth were related to those in precipitation NPP to total NPP ranged from 0.25-0.6 for semi-arid and grazing. grasslands of North America (Sims and Singh, 1978; The field measurements at Bayan Unjuul for 2005- Olson et al., 2001; Milchunas and Lauenroth, 1992). 2007 have shown that AGP was significantly higher In fact, BGP is several times AGP for Bayan Unjuul in no grazing plots than in grazing plots in 2005 and (Shinoda et al., 2009), corresponding to a few times 2006, whereas the opposite result was obtained in annual belowground NPP. This fact implies evidence of 2007 (but not significant). It is therefore likely that the carryover of BGP from one year to another in the grazing effects might be under the control of pasture Mongolian perennial grasslands. Belowground NPP has availability to livestock. As for 2007, there appeared often been estimated based on the ratio of aboveground to be less or no consumption of phytomass due to NPP to belowground NPP rather than directly due low pasture availability in the steppe region; hence, to the difficulties associated with belowground NPP herders moved their livestock to regions having more measurement. Therefore, a predefined ratio, as obtained pasture. in this study, will be valuable to estimate total NPP Potter et al. (1993) simulated annual NPP using the of natural grasslands. CASA model and it was estimated as 28 gC m–2 yr–1 In the present investigation, simulation exercise was for desert and 180 gC m–2 yr–1 for perennial grasslands. implemented year by year. However, several studies On the other hand, the Century model estimated have pointed out that drought, as a disturbance, had annual NPP of 27 gC m–2 yr–1 for the desert, 100 negative impacts on plant production in subsequent gC m–2 yr–1 for the steppe, and 290 gC m–2 yr–1 for years (Haddad et al., 2002; Lauenroth and Sala, 1992; the forest of Inner Mongolia and Mongolia (Chuluun Oesterheld et al., 2001; Wiegand et al., 2004). On and Ojima, 2002)). Our simulated NPP values in the the other hand, as for Bayan Unjuul, AGP recovered Mongolian grasslands were substantially lower than quickly in the following year of the drought experi- those estimated by the above studies; this may be ment, likely because BGP (which was several times because our study years were among the driest years AGP) was not severely damaged by the drought during 1995-2007. (Shinoda et al., 2009). There is no such information The amount of NPP allocated below the ground for the other stations due to no BGP data available. remains among the most poorly understood attributes From the above-mentioned results, we obtained the of ecosystems (Lauenroth, 2000), and very few existing following main conclusions: studies have obtained direct measurements of BGP. The modified CASA model simulated, reasonably In most cases, BGP has been ignored, leading to well, the spatial variation of NPP in the Mongolian underestimation of carbon fluxes (Long et al., 1989, grasslands. The simulated NPP differed substantially 1992). For these reasons, it is important to explore the from one vegetation region to another. Further model dynamics of root system in order to reach an accurate modification, calibration, and validation for more sites assessment of NPP in grasslands. and years will be essential for future practical use of Belowground primary production is often greater the model to support the grassland management. than aboveground production in the perennial na- The analysis of effects of temperature and water tive ecosystems of arid and semi-arid grasslands stresses on pasture productivity demonstrated that (Coleman, 1976; Sims and Singh, 1978) and over a water stress is stronger down-regulator of NPP in the large proportion of the world's land area (Coupland, Mongolian grasslands. The water balance submodel of 1992). The present study showed that measured AGP the CASA model improved by Nemoto et al. (2003) (namely, a good approximation of aboveground NPP successfully represented water stress for both the dry for the grassland ecosystem) was about 0.4 times and wet years. the simulated total NPP over the three years in the Comparison of simulated NPP and measured AGP Mongolian perennial grasslands. Since as mentioned indicated that aboveground NPP:belowground NPP in Section 2.2, the animal-available AGP used for the is 1：1.5 in the Mongolian perennial grasslands (40％ three sites is a slightly underestimate of the total AGP, for aboveground and 60％ for belowground). This this ratio may be slightly larger than 0.4. In either case, ratio falls within the range reported by the previous the ratio is likely to fall within the range indicated studies. by the previous studies; the ratio of aboveground

37 J. Agric. Meteorol. (農業気象) 66 (1), 2010

Environ., 51, 74–88. Acknowledgments Goward, S. N., and Huemmrich, K. F., 1992: Vegetation The authors would like to thank the Institute of canopy PAR absorptance and the normalized differ- Meteorology and Hydrology of Mongolia for providing ence vegetation index: An assessment using the SAIL the required data. Also, we are grateful to the Land model. Remote Sens. Environ., 39, 119–140. Processes Distributed Active Archive Center (LP Gunin, P. D., Vostokova, E. A., Dorofeyuk, N. I., DAAC) and the Earth Observing System Data Gateway Tarsov, P. E., and Black, C. C., 1999: Natural and for providing the MODIS NDVI data. This research anthropogenic factors and the dynamics of vegetation was supported by Grants-in-Aid for Scientific Research distribution. In Vegetation dynamics of Mongolia from the Japanese Ministry of Education, Science, (ed. by Werger, M. J. A.), Geobotany, 26, Kluwer Sports, and Culture (Nos. 16405002, 20255001). Academic Publishers, AH Dordrecht, Netherlands, pp. 7–43. References Haddad, N. M., Tilman, D., and Knops, J. M. H., 2002: Batima, P., and Dagvadorj, D., 2000: Climate Change Long-term oscillations in grassland productivity in- and Its Impact in Mongolia. JEMR Publishing, duced by drought. Ecol. Lett. 5, 110–120. , pp. 46–93. Johnson, D., Sheehy, D., Miller, D., and Damiran, D., Bolortsetseg, B., Erdenetsetseg, B., and Bat-Oyun, 2006: Mongolian rangelands in transition. Sécher- Ts., 2002: Impact of past 40 years climate change esse, 17 (1–2), 133–141. on pasture plant phenology and production. Papers Kondoh, A., and Kaihotsu, I., 2003: Preliminary analy- in Meteorology and Hydrology of Mongolia, 24, sis on the relationship between vegetation activity 108–114. and climatic variation in Mongolia (in Japanese). J. Bolortsetseg, B., 2006: Climate change Vulnerability Arid Land Stud. 13 (2), 147–151. and Adaptation in the Livestock Sector of Mongolia. Lauenroth, W. K., and Sala, O. E., 1992: Long-term Assessments of Impacts and Adaptations to Climate forage production of North American shortgrass Change (AIACC) final report. Ulaanbaatar, 44 pp. steppe. Ecol. Appl., 2, 397–403. Chen, Y., Lee, G., Lee, P., and Oikawa, T., 2007: Lauenroth, W. K., 2000: Methods of estimating be- Model analysis of grazing effect on above-ground lowground net primary production. In Methods in biomass and above-ground net primary production ecosystem ecology (ed. by Sala, O. E., Jackson, R. of a Mongolian grassland ecosystem. J. Hydrol., B., Mooney H. A., and Howarth R. W.), Springer, 333, 155–164. New York, pp. 58–71. Chuluun, T., and Ojima, D., 2002: Land use change Lobell, D. B., Hicke, J. A., Asner, G. P., Field, C. B., and carbon cycle in arid and semi-arid lands of East Tucker, C. J., and Los, O., 2002: Satellite estimates and Central Asia. Sci. China., 45, 48–54. of productivity and light use efficiency in United Coleman, D. C., 1976: A review of root production States agriculture, 1982–98. Glob. Change Biol., 8, processes and their influence on soil biota in ter- 722–735. restrial ecosystems. In The Role of Terrestrial and Long, S. P., Garcia Moya, E., Imbamba, S. K., Kamn- Aquatic Organisms in Decomposition Processes. alrut A., Piedade, M. T. F, Scurlock, J. M. O, Shen, (ed. by Anderson, J. M., and Macfadyen, A.), Y. K., and Hall, D. O., 1989: Primary productivity Blackwell Scientific Publications, Oxford, England, of natural grass ecosystems of the tropics: a reap- pp. 417–434. praisal. Plant Soil, 115, 155–166. Coupland, R. T., 1992: Mixed prairie. In Grasslands Long, S. P., Jones, M. B., and Roberts, M. J., (eds.) of the world (ed. by Coupland, R. T.), Cambridge 1992: Primary Productivity of Grass Ecosystems University Press, Cambridge, pp. 151–182. of the Tropics and Sub-tropics. Chapman and Hall, Fernandez-Gimenez, M. E., and Allen-Diaz, B., 1999: London. Testing a non-equilibrium model of rangeland veg- Milchunas, D. G., and Lauenroth, W. K., 1992: Carbon etation dynamics in Mongolia. J. Appl. Ecol., 36, dynamics and estimates of primary production by 871–885. harvest, 14C dilution, and 14C turnover. Ecology, Field, C. B., 1995: Global net primary production: 73 (2), 593–607. combining ecology and remote sensing. Remote Sens. Miyazaki, S., Yasunari, T., Miyamoto, T., Kaihotsu, I., Davaa, G., Oyunbaatar, D., Natsagdorj, L., and Oki,

38 T. Bat-Oyun et al. : Estimation of pasture productivity in Mongolian grasslands

T., 2004: Agrometeorological conditions of grassland America: Application of a global model. Ecol. Appl., vegetation in central Mongolia and their impact for 1, 399–429. leaf area growth. J. Geophys. Res., 109, D22106. Sellers, P. J., 1987: Canopy reflectance, photosynthe- Monteith, J. L., 1972: Solar radiation and productivity sis, and transpiration. II. The role of biophysics in in tropical ecosystems. J. Appl. Ecol., 9, 747–766. the linearity of their interdependence. Remote Sens. Monteith, J. L., 1977: Climate and efficiency of crop Environ., 21, 143–183. production in Britain. Philos. Trans. R. Soc. B-Biol. Sellers, P. J., Berry, J. A., Collatz, G. J., Field, C. B., Sci., 281, 271–294. and Hall, F. G., 1992: Canopy reflectance, photo- Munkhtsetseg, E., Kimura, R., Wang, J., and Shinoda, synthesis and transpiration, III, A reanalysis using M., 2007: Pasture yield response to precipitation improved leaf models and new canopy integration and high temperature in Mongolia. J. Arid Environ., scheme. Remote Sens. Environ., 42, 187–216. 70, 94–110. Shinoda, M., Ito, S., Nachinshonhor, G. U., and - Nachinshonhor, G. U., 2001: Influence of climate and setseg, D., 2007: Phenology of Mongolian grasslands nomadic activities on the plant community of Mon- and moisture conditions. J. Meteorol. Soc. Jpn., 85 golian grasslands. PhD thesis, Tohoku University, (3), 359–367. 207p Shinoda, M., Nachinshonhor, G. U., and Nemoto, M., Nakano, T., Nemoto, M., and Shinoda, M., 2008: En- 2009: Impact of drought on vegetation dynamics of vironmental controls on photosynthetic production the Mongolian steppe: A field experiment. J. Arid and ecosystem respiration in semi-arid grasslands of Environ., in press. Mongolia. Agric. For. Meteorol., 148, 1456–1466. Sims, P. L., and Singh, J. S., 1978: The structure Nemoto, M., Shinoda, M., and Ju, H., 2003: Developing and the function of ten western north American a simple soil Moisture model for semi-arid regions: grasslands. III. Net primary production. Turnover A case for Ordos in Inner Mongolia. J. Agric. Me- and efficiencies of energy capture and water use. teorol., 59(1), 51–58. J. Ecol., 66, 573–597. Ni, J., 2003: Plant functional types and climate along Suzuki, R., Masuda, K., and Dye, D. G., 2007: Interan- a precipitation gradient in temperate grasslands, nual covariability between actual evapotranspiration north-east China and south-east Mongolia. J. Arid and PAL and GIMMS NDVIs of northern Asia. Environ., 53, 501–516. Remote Sens. Environ., 106, 387–398. Oesterheld, M., Loreti, J., Semmartin, M., and Sala, O. Tao, F., Yokozawa, M., Zhang, Z., Xu, Y., and Hayashi, E., 2001: Inter-annual variation in primary produc- Y., 2005: Remote sensing of crop production in tion of a semi-arid grassland related to previous-year China by production efficiency models: models com- production. J. Veg. Sci., 12, 137–142. parisons, estimates and uncertainties. Ecol. Model., Olson, R. J., Johnson, K. R., Zheng, D. L., and Scur- 183, 385–396. lock, J. M. O., 2001: Global and regional Ecosystem Thornthwaite, C. W., 1948: An approach towards a Modeling: Databases of model drivers and valida- rational classification of climate. Geogr. Rev., 38, tion measurements. ORNL, Technical Memorandum 55–94. TM–2001/196, Oak Ridge National Laboratory, Oak Wiegand, T., Snyman, H. A., and Kellner, K., 2004: Ridge, Tennessee, U.S.A. Do grasslands have a memory: modeling phytomass Potter, C., Klooster, S., Tan, P., Steinbach, M., Kumar, production of a semiarid South African grassland. V., and Genovese, V., 2003: Variability in terres- Ecosystems., 7, 243–258. trial carbon sinks over two decades: Part 1. North Yuan, J., Niu, Z., and Wang, C., 2006: Vegetation America, Earth Interactions, 7 Paper 12. NPP Distribution Based on MODIS Data and CASA Potter, C. S., Randerson, J. T., Field, C. B., Matson, P. Model-A Case Study of Northern Hebei Province. A., Vitousek, P. M., Mooney, H. A., and Klooster, S. Chin. Geogr. Sci., 16(4), 334–341. A., 1993: Terrestrial ecosystem production: a process Zhang, Y., Munkhtsetseg, E., Kadota, T., and Ohata, model based on global satellite and surface data. T., 2005: An observational study of ecohydrology Glob. Biogeochem. Cycle, 7, 811–841. of a sparse grassland at the edge of the Eurasian Raich, J. W., Rastetter, E. B., Mellilo, J. M., Kick- cryosphere in Mongolia. J. Geophys. Res., 110, lighter, D. W., Steudler P. A., and Peterson, B. J, D14103. 1991: Potential net primary productivity in South

39

View publication stats