Forage Supply–Demand Assessment and Influencing Factor Analysis from the Perspective of Socio-Ecological System: A Case Study of Altay Prefecture, China

Abstract

The provision and utilization of grassland resources connect grassland ecosystems and pastoral society. Revealing the mechanism behind the forage supply–demand relationship and balancing forage supply and demand is essential in pastoral socio-ecological systems. Taking Altay Prefecture as the case study, this study quantified the dynamics of natural forage supply, artificial supplemental forage, and forage demand. The ratio of forage supply to demand was calculated in the traditional grazing scenario and the grazing prohibition and supplemental feeding scenario. The results showed that during 2001–2018, the forage supplied by natural grasslands fluctuated, with the highest and lowest values in 2013 and 2008, respectively. The artificial supplemental forage increased at a higher rate in 2011–2018 than in 2001–2010. The overall trend of forage demand was upward, at approximately 2.98 × 10⁴ t/a. The ratio of forage supply to demand decreased in the traditional scenario with an average value of 0.4717 and increased in the grazing prohibition and supplemental feeding scenario with an average value of 0.8289. The influencing factors were analyzed from the perspective of the interactions between the socio-ecological system elements, and the forage supply–demand relationships were conceptualized based on the social–ecological system framework. This study concludes that natural grasslands cannot entirely meet the increasing forage demand without artificial supplemental forage. The increasing artificial supplemental forage has promoted a balance between forage supply and demand, indicating an optimized grassland utilization pattern. The factors that affect forage supply–demand relationships are interrelated, and a holistic perspective should be adopted when implementing management measures.

1. Introduction

Grasslands cover approximately 40% of the Earth’s surface, and their area is about 8.89 million km² in Asia [1]. In China, grassland ecosystems are an important ecosystem type, occupying more than 40% of terrestrial land [2]. Grasslands are multifunctional, including the ecological function that supports lives and regulates the environment, the production function that provides the basic living materials such as forage and livestock products, and the living function such as habitation [3]. Grasslands can provide various ecosystem services, including provisioning, regulating, cultural, and supporting services, and contribute to human wellbeing [4]. Provisioning services are closely related to the subsistence of pastoral people, among which forage production is one of the most important because grazing activities largely rely on the forage provided by grasslands [5,6,7]. However, it has been reported that China’s grasslands have undergone marked degradation in recent years [8], threatening the sustainable supply of forage and challenging the sustainability of pastoral regions [9]. One of the most intuitive manifestations is the imbalance between forage supply and demand. Hence, coordinating the relationship between forage supply and demand has been regarded as an effective means of protecting grasslands and improving pastoral livelihoods [10].

In recent years, the concept of the socio-ecological system (SES), emphasizing the concept of society and nature as an interlinked whole, has provided new and helpful perspectives to address sustainability challenges [11,12]. The SES has been regarded as a complex system consisting of social and ecological subsystems and a series of feedback and interactions despite the differences in the definitions [13]. In the pastoral natural resource utilization system, forage, livestock, and herders are interrelated [14]; the delivery of forage from the supply side of grasslands to the demand side of pastoral societies connects the social and ecological subsystems of pastoral SESs [15,16]. The interactions between the elements of pastoral SESs have shaped local ecological and socioeconomic conditions and have also been shaped by broader or external factors. Such complex interactions may be the potential mechanism behind forage supply–demand relationships [17,18]. Therefore, quantifying the dynamic relationship between forage supply and demand and analyzing the influencing factors are meaningful for revealing the interactions of the elements of pastoral SESs and for developing management measures.

The forage supply–demand relationship is a hot topic in research fields such as ecosystem services, grassland carrying-capacity assessment, and grassland ecosystem management [5,19]. Studies concerning quantifying forage supply and demand have been carried out across different spatial and temporal scales. For quantifying the forage supply, scholars have often used the grassland aboveground biomass (AGB) as the indicator [20]. The grassland AGB can be estimated by remote sensing methods based on, for example, the grassland net primary productivity (NPP) or normalized difference vegetation index at the large scale [21,22], field experiments based on quadrat surveys at the local scale [23] or a combination of both [24]. The indicator regarding the forage quality, such as the content of nutrients and crude protein, has also been selected by many to value forage supply, and these studies have mainly been carried out locally [9,25]. To quantify forage demand, scholars usually use parameters such as livestock number, slaughter rate, daily feed, and grazing days [10]. Forage supply–demand relationships are often quantified from two aspects. One is from the aspect of the livestock number, employing the ratio of the actual grazing livestock number to the theoretical carrying capacity of grasslands; that is, the grazing pressure index, to measure the overgrazing status of grasslands [26]. The other is from the aspect of the quantity of forage, using the ratio of the actual forage consumption to the total available forage as the indicator [27]. In addition, some studies have taken into account the forage intake of wild herbivores [28] when calculating forage supply–demand relationships. Some studies have also reported the intake rate and forage availability of small ruminants such as goats [29,30].

Previous studies have identified several factors influencing forage supply and demand dynamics. For forage supply, climate change and human activities are the main influencing factors [31]. The variation in precipitation and temperature can affect grassland growing conditions [26]. Extreme heat and drought events during the growing season will decrease the forage supply of natural grasslands and the yield of artificial grasslands. Human activities, such as overgrazing, can cause grassland degradation and forage decreases [32]. The dynamics of forage demand are closely related to the livestock number, which could be driven by the combined effects of policies, ecological construction, and natural factors, including severe climate and disease [33]. The interactions of the ecological and social subsystems are complex, bringing difficulties in applying a holistic perspective in analyzing the forage supply–demand relationship. Recent advances in the social–ecological system framework (SESF) [34,35] have provided a new point for understanding and analyzing the interactions in pastoral SESs and have been applied in studies, e.g., changes in institutions and policies in grassland management [36], ecosystem services payments in the Eurasian steppe [37], and vulnerability analysis of pastoral systems [38]. However, research on forage supply–demand from the perspective of social–ecological interactions is relatively scarce.

Altay Prefecture is a typical pastoral SES located in inland Eurasia, and the grasslands in Altay Prefecture have experienced overgrazing in past decades [39]. Many management policies (such as grazing prohibition and restriction) and measures (such as grassland seeding and improvement) have been implemented at different levels to promote grassland restoration [40,41]. According to a previous study [42] and the information acquired from local authorities, the development of animal husbandry in Altay Prefecture has gone through different stages. After the 2000s, with the implementation of relevant policies and the development of economics, traditional transhumant pastoralism transformed into a modern pattern characterized by grazing in the warm season and house feeding in the cold season. Under this transition, the relationship between forage supply and demand is undergoing new changes. However, how these changes occurred and what effects will be caused by these changes are unclear, which may impose some barriers on grassland resource management and confusion in understanding the interactions in pastoral SESs.

This study took Altay Prefecture as the case study to explore the dynamics and influencing factors of the forage supply, demand, and supply–demand relationship. The SESF was applied in this study to group the influencing factors into different SES elements for a better understanding of the impacts of the interactions of SES elements on forage supply–demand dynamics. The main contents of this study included (1) investigating the dynamics of the forage supply, demand, and supply–demand relationships and (2) revealing the interactions in pastoral SES that drive the forage dynamics. The contributions of this study include providing an understanding of the forage supply and demand dynamics from the SES perspective and, on this basis, providing implications for grassland management and sustainability in pastoral regions.

2. Materials and Methods

2.1. Study Area

Altay Prefecture is located in Northwest China with an area of approximately 11.8 × 10⁴ km², and it borders Kazakhstan, Russia, and Mongolia. Located in the interior of Eurasia, Altay Prefecture has a temperate continental climate; the annual precipitation was 193.45–338.36 mm, and the annual mean temperature was 2.01–4.32 °C from 2001 to 2018 [43]. The monthly precipitation and temperature in January were about 13.01 mm and −18.47 °C, and the values were about 37.88 mm and 20.65 °C in July. The Altai Mountains lie in the northern part of Altay Prefecture, the Sawuer Mountains are in the southwest, and the vast Junggar Basin is in the middle and south of Altay Prefecture. The altitude of Altay Prefecture decreases from north to south, ranging between 366 m and 4168 m. The main rivers include the Irtysh River and the inland Ulungur River (Figure 1a). Under the combined effects of the abovementioned climatic and terrain characteristics, the grassland ecosystem is the dominant ecosystem in the study area, occupying more than 80% of the total area (Figure 1b). These grasslands are usually used as grazing pastures and are mainly classified into three seasonal pastures: summer pasture, spring-autumn pasture, and winter pasture. There is also a small proportion of grasslands that are used for three seasons or the whole year.

Altay Prefecture is one of China’s most important pastoral areas, and animal husbandry is an important industry [44]. According to the statistical yearbook [45], the gross output value of animal husbandry was 35.41 × 10⁸ CNY in 2018, which accounted for approximately 42.13% of the gross output value of agriculture. The total population was 65.95 × 10⁴ in 2018, and the number of rural households was 9.75 × 10⁴, of which approximately one-third were pastoral households. In different seasons, the herdsmen in Altay Prefecture traditionally graze their livestock in corresponding seasonal pastures: from mid-June to mid-September, they spend approximately 90 d grazing in summer pastures located in mountain areas; from September to early December, they graze in spring-autumn pastures for approximately 85 d; from December to March, they graze in winter pastures located in the Junggar Basin for approximately 120 d; and from late March to early June, they spend approximately 70 d grazing in spring-autumn pastures, in which the spring-autumn pastures are usually utilized twice a year in spring and autumn. Altay Prefecture has one city and six counties, namely, Altay, Burqin, Fuyun, Fuhai, Habahe, Qinghe, and Jeminay (Figure 1a). The changes in the number and structure of livestock (mainly sheep, goats, cattle, horses, and camels) in Altay Prefecture from 2001 to 2018 are shown in Figure 2. It can be found that the overall trends of the number of both slaughtered livestock and year-end livestock increased from 2001 to 2018. The multiyear average numbers of the slaughtered livestock and year-end livestock were 296.41 × 10⁴ SU and 573.73 × 10⁴ SU, respectively. The number of slaughtered livestock increased by 28.86% from 2001 to 2018, while that of the year-end livestock increased by 14.41%.

The Chinese government has implemented a series of policies or ecological construction projects to restore degraded grasslands and promote the sustainability of pastoral regions. They include the Returning Grazing Land to Grassland (RGLG), the Herdsmen Settlement Project (HSP), the Subsidy and Incentive System for Grassland Conservation (SISGC), and the Forage–Livestock Balance (FLB) [46] (

Table 1. Major policies and projects implemented after 2000.

Policies and Projects	Starting Time	Main Aim	Main Measures	References
RGLG	2003	To rest and restore degraded grasslands and reduce grazing pressure to promote the sustainability of grassland resources.	Controlling livestock numbers to prevent overgrazing by grazing prohibition, grazing restriction, and rotational grazing; subsidizing grassland fence construction, fodder, and grain.	[40,41]
FLB	2005	Achieve a dynamic balance between the available forage from natural grasslands and other sources and the livestock demand.	Encouraging herdsmen to plant artificial forage, purchase extra forage, increase the slaughter rate, and stall-feeding livestock to approach forage–livestock balance.	[47]
HSP	2009	To improve pastoral livelihoods by increasing living standards, income, and quality of life and to protect grasslands by upgrading animal husbandry after settlement.	Building standardized houses, yards, and sheds; reclaiming artificial grasslands; ensuring the access of the settlements to electricity, water, and roads; building infrastructure such as hospitals, schools, and cultural stations.	[41,46]
SISGC	2011	Mitigate grassland ecological deterioration; optimize animal husbandry development and strengthen pastoral economy sustainability; increase herdsmen’s income.	Compensating herdsmen who stop grazing in prohibited areas (including grasslands that were severely degraded and highly valued in water conservation and grassland protected areas); rewarding herdsmen who reduced their grazed livestock according to grassland productivity; subsidizing for production machinery, fine grass seeds, and livestock breeds.	[40,47,48]

Note: in the first column, RGLG means Returning Grazing Land to Grassland, FLB means Forage-Livestock Balance, HSP means Herdsmen Settlement Project, and SISGC means Subsidy and Incentive System for Grassland Conservation.

).

2.2. Data and Processing

The data used in this study mainly included spatial datasets and statistical data. Yearly NPP data with a spatial resolution of approximately 500 m from 2001 to 2018 were obtained from the National Earth System Science Data Center (http://www.geodata.cn (accessed on 1 November 2021)) [49] and were used to calculate the grassland AGB. The vector data on grassland resource types in Xinjiang (1:100 million) were acquired from the National Cryosphere Desert Data Center (http://www.ncdc.ac.cn (accessed on 28 November 2021)). The map of seasonal pasture boundaries (1:50 million) was obtained from the Grassland Ecology Station of Altay Prefecture and was digitized to vector format. The data of measured grassland AGB in July at fixed monitoring sites from 2014 to 2016 were obtained from the Grassland Ecology Station of Fuyun County. The sites in the natural grazing pastures were selected (Figure 1b) and used to validate the estimated AGB. The temperature and precipitation data with a time interval of 8 d and a spatial resolution of 1 km between 2001 and 2018 were acquired from the National Ecosystem Science Data Center (http://www.nesdc.org.cn/ (accessed on 17 December 2021)) [50]. The statistics, including the number and types of livestock and the area and yield of the crops, were obtained from the province-level, city-level, and county-level statistical yearbooks, and they were used for quantifying the artificial supplemental forage and the forage demand. Geographic information data were obtained from the National Catalogue Service for Geographic Information (https://www.webmap.cn (accessed on 15 March 2021)). Other data regarding the spatial and temporal characteristics of grazing activities were obtained from field surveys. All spatial datasets were unified into a raster format with the same coordinate system (WGS84_Albers) and the same spatial resolution (1 km) using ArcGIS 10.2. The statistical data were processed and analyzed using Microsoft Excel 2016 and Origin 2022.

2.3. Methods

2.3.1. Quantifying Forage Supply

The forage supply includes the forage provided by natural grasslands (FSn) and artificial supplemental forage (FSa). The FSn was calculated following the studies of grassland carrying-capacity assessment [10,27], and the first step was to calculate the grassland AGB based on NPP data:

$$Y_i=\frac{NPP}{t\left(1+r\right)}$$

(1)

where Y_i is the grassland AGB at location i (g/m²); NPP is the net primary productivity at location i (g C/m²); t is the conversion coefficient of biomass to NPP and was set as 0.45 [10]; r is the ratio of belowground biomass to aboveground biomass, and the determination of r was based on a previous study [51]. The information on r can be found in

Table 2. The parameters for calculating the forage supplied by natural grasslands.

Grassland Types	r	Co/%	Ut/%
Temperate meadow steppe	5.26	89	65
Temperate steppe	4.25	81	65
Temperate desert steppe	7.89	81	60
Alpine steppe	4.25	91	45
Temperate steppe desert	7.89	75	45
Temperate desert	7.89	67	40
Lowland meadow	6.31	81	55
Mountain meadow	6.23	97	65
Alpine meadow	7.92	87	65
Swamp	15.68	62	55

Note: r is the ratio of belowground biomass to aboveground biomass, C_o is the area factor (%), and U_t is the grassland utilization rate (%).

.

The total AGB of natural grasslands can only partially be eaten by livestock because there are poisonous and unpalatable parts, and inaccessible parts limited by steep terrain and grazing time. Therefore, the second step was to revise the Y_i to calculate the available forage from natural grasslands [52,53]:

$$FS{n}_i=Y_i\times C_o\times U_t\times H_a$$

(2)

where FSn_i is the forage provided by natural grasslands at location i (g/m²); C_o is the area factor (%), which was calculated as the ratio of the available grassland area to the total grassland area [54]; U_t is the grassland utilization rate (%), which means the ratio of available biomass to total aboveground biomass, and was determined based on the information from the field survey and the study of [55]. H_a is the percentage of edible grasses (%) [53]. C_o and U_t were determined based on grassland types (Table 2), while H_a was determined based on seasonal pasture types, and the values for the summer, winter, and spring-autumn pastures were 0.90, 0.80, and 0.85, respectively [52]. Finally, the total FSn (10⁴ t) was calculated by the sum of the products of FSn_i and the corresponding pixel area.

The FSa in Altay Prefecture includes alfalfa, silage corn, grain corn, and crop straw. According to the field survey, the crop straw mainly consists of wheat stalks and corn stalks. The FSa was calculated as follows:

$$FSa=\sum FS{a}_i$$

(3)

where FSa is the total artificial supplemental forage (10⁴ t); FSa_i is the forage yield from different sources (10⁴ t); and i is the type of artificial supplemental forage. In addition to these four types of FSa, the forage harvested from natural-mowing pastures was one type of FSa that was not counted because these pastures are mainly lowland meadows, and this part of the forage has been included within the FSn. For years, when the yield of FSa_i was missing, we used the product of the planted area in that year and the average yield per unit area in other years. The yields of wheat stalks and corn stalks were estimated based on the ratio of straw to grain, and the ratios were 1.1 and 1.2 for wheat and corn, respectively [56].

In northern Xinjiang, due to the uneven areas of grassland types and seasonal pastures in different counties, herdsmen in one county may graze in another county in different seasons based on historical grazing habits. In addition, the grasslands in certain areas may be inaccessible due to the implementation of grazing prohibition. Therefore, in addition to the FSn and FSa, the forage acquired from cross-region grazing areas (FSc) needs to be added, and the inaccessible forage (FSp) in grazing-prohibition areas needs to be subtracted. The FSc and FSp were calculated as follows.

$$FSc=C{A}_i\times FS{n}_i$$

(4)

$$FSp=P{A}_i\times FS{n}_i$$

(5)

where CA_i is the area of the cross-region pasture (km²), PA_i is the area of the grazing prohibition pasture (km²), i is the type of seasonal pasture, and FSn_i is the per-unit forage production of the i-th type of seasonal pasture in Altay Prefecture (g/m²). Cross-region grazing pastures are mainly winter pastures, and grazing prohibition pastures are mainly winter and summer pastures.

2.3.2. Quantifying Forage Demand

The forage demand was calculated based on the number of livestock, daily feed, and grazing days. The calculation of FD was as follows:

$$FD=C_i\times S\times D\times G_i+C_s\times S\times D\times G_s$$

(6)

where FD is the total demand of forage (10⁴ t); C_i is the livestock number at year end (sheep unit, SU); S is the daily feed per SU (kg), set to 4 kg of fresh grass per day; D is the dry weight ratio to the fresh weight of forage, set to 1:3 [10]; G_i is the grazing time of inventory livestock, which was set as 365 d; C_s is the number of slaughtered livestock (SU); and G_s is the grazing time of the slaughtered livestock, which was set as 180 d [27]; All types of livestock were converted into SU, and the camel, horse, cattle, goat, and sheep values equated to 7, 6, 5, 0.8, and 1 SU, respectively.

2.3.3. Quantifying Forage Supply–Demand Relationship

This study employed the ratio of forage supply to demand (RSD) to represent the forage supply–demand relationship. Considering the actual context, we defined two scenarios to compare the differences between RSD without and with management interventions. Scenario I is the traditional grazing scenario in which the herdsmen graze in all seasonal pastures following the historical grazing habits without any management interventions. Scenario II is the grazing prohibition and supplemental feeding scenario in which the herdsmen can graze in pastures outside the area of grazing-prohibition projects and supplement artificial forage to their livestock. Grazing-prohibition projects were implemented not only in Altay Prefecture but also in the surrounding administrative regions, and grazing has been prohibited in many of the abovementioned cross-region grazing areas. Therefore, in scenario II, the forage sourced from the cross-region grazing areas was excluded. The RSD in scenario I and scenario II was calculated as follows.

$$RS{D}_I=\frac{FSn+FSc}{FD}$$

(7)

$$RS{D}_{II}=\frac{FSn-FSp+FSa}{FD}$$

(8)

where RSD_I and RSD_II are the ratios of forage supply to demand in scenario I and scenario II, respectively; FSn is the forage supplied by natural grasslands in Altay Prefecture (10⁴ t); FSc is the forage obtained from cross-region grazing (10⁴ t); FSp is the forage in grazing prohibition regions that is unacquirable (10⁴ t); FSa is the artificial supplemental forage (10⁴ t); and FD is the total forage demand (10⁴ t).

2.3.4. Statistical Analysis

This study used the partial correlation analysis method to explore the relationship between climate factors and FSn. The climate factors included the growing-season precipitation and average temperature, and the growing seasons were defined as spanning from April to September. The partial correlation coefficients were calculated following the study of Guo et al. [57].

3. Results

3.1. Forage Supply

3.1.1. Validation of Grassland AGB

The modeled results of grassland AGB were compared with the field-measured data to verify the accuracy. Figure 3a shows the correlation between the observed and estimated grassland AGB in three years (R² = 0.6229, p < 0.01), and Figure 3b illustrates the correlation between the three-year averaged observed and estimated grassland AGB (R² = 0.7268, p < 0.01), indicating a reliable estimation accuracy. The estimated grassland AGB in mountain areas, such as the alpine meadow, was often higher than the observed grassland AGB; conversely, the estimated grassland AGB in plain areas, such as temperate desert, was often lower than the observed grassland AGB. The possible reason for these differences is that the vegetation types in mountain areas are abundant, and the grassland pixels may be mixed pixels of grasslands and shrubs or forests, resulting in an overestimation. The grassland pixels in plain areas may be mixed pixels of grasslands and bare lands, causing an underestimation.

3.1.2. Spatial and Temporal Characteristics of FSn

Figure 4 displays the spatial distribution of FSn per unit area during 2001–2018. The FSn in Altay Prefecture had apparent characteristics of change along the altitude gradients. The FSn decreased from the northwest to the southeast and was high in mountain areas and low in desert areas. In the middle and western sections of the Altai Mountains and parts of the Sawuer Mountains, the FSn was generally greater than 50 g/m². In the eastern section of the Altai Mountains and much of the Sawuer Mountains, the FSn was mainly between 10 and 50 g/m². The FSn within this value range was also found in the low mountains of the Western and Middle Altai Mountains. The area with the highest FSn occurred northwest of the Altai Mountains, exceeding 100 g/m², mainly because of the abundant precipitation in mountain areas, especially in the western part of the Altai Mountains. The FSn was also high in the riparian areas of the Irtysh River and the Ulungur River, mainly due to the close distance to the rivers. In the areas between the two rivers and some parts south of Ulungur Lake, the FSn was mainly between 2.5 and 5 g/m². In the southern part of the study area, the FSn was mostly lower than 2.5 g/m².

As shown in Figure 5a, the average value of the FSn per unit area was 15.25 g/m² and there was no significant upward or downward trend during 2001–2018. The FSn per unit area increased and decreased alternately between 2001 and 2005 and continuously decreased from 2005 to 2008. In 2008, the FSn per-unit area was the lowest during the study period, with a value of 12.27 g/m². From 2008 to 2013, the FSn per unit area increased rapidly during 2008–2009 and 2012–2013, reaching the highest value of 18.92 g/m² in 2013. From 2013 to 2014, the FSn per unit area decreased rapidly and then fluctuated after 2014. The temporal trend of the total FSn was the same as that of the FSn per unit (Figure 5b), with an average value of 154.42 × 10⁴ t. The lowest FSn was in 2008, with a value of 124.20 × 10⁴ t, and the highest FSn was 191.55 × 10⁴ t in 2013.

3.1.3. The Partial Correlations between the FSn and Climate Factors

This study used pixel-based partial correlation analysis to explore the relationship between FSn and precipitation and the average temperature of the growing season. Figure 6a shows that the partial correlation between the FSn and growing-season precipitation was mainly positive. The number of positively correlated pixels accounted for 85.6% of the total pixels, of which 24.20% were significant, indicating that the higher the precipitation in the growing season was, the greater the FSn. Spatially, it can be found that the pixels, with a significantly positive correlation with growing-season precipitation, were mainly distributed in the low mountains. Conversely, the partial correlation between the FSn and growing-season average temperature was mainly negative (Figure 6b), with negatively correlated pixels accounting for 73.63% of the total pixels, of which 19.67% were significant. The pixels that negatively correlated with the growing-season average temperature were mainly distributed in mountain areas.

3.1.4. Temporal Dynamics of FSa

The dynamics of artificial supplemental-forage production are shown in Figure 7. As plotted in Figure 7a, from 2001 to 2018, the FSa of Altay Prefecture continuously increased except in 2007–2008 and 2011–2012. The average FSa from 2001 to 2018 was 145.06 × 10⁴ t; moreover, the lowest and highest FSa values were in 2001 and 2018, with values of 57.60 × 10⁴ t and 387.53 × 10⁴ t, respectively. From 2001 to 2018, the FSa increased by approximately 572.80%. The rate of increase of FSa was low during 2001–2010, with a rate of approximately 6.62 × 10⁴ t/a, and high after 2010, with a rate of approximately 34.64 × 10⁴ t/a. In the composition of different FSa types in Altay Prefecture (Figure 7b), the multiyear average proportions of the four types of FSa, from high to low, were silage corn (42.77%) > crop straw (24.40%) > alfalfa (20.53%) > corn (12.30%). In 2005–2011 and 2016–2018, the proportion of silage corn was high, occupying nearly or more than half of the total production. The proportion of alfalfa generally decreased from 41.3% to 9.7%. Silage corn and alfalfa were the main types of artificial grasslands, and they provided forage with high quality and high production per unit area and were the primary sources for supplemental feeding. The average proportion of these two types was 63.31% from 2001–2018, and the value increased to approximately 80% between 2006–2007 and 2016–2018.

A comparison of Figure 5a and Figure 7a shows that the temporal dynamics of FSn and FSa have different variation trends. The overall trend of the FSn fluctuated between 2001 and 2018, while that of the FSa increased. Figure 8 compares the proportions of FSn and FSa from 2001 to 2018. In 2001, the FSn occupied approximately 73.5% of the total FS and was approximately three times the FSa. From 2001 to 2011, the proportion of FSa eventually increased to approximately 50%. The proportion of FSa began to exceed that of FSn after 2014, reaching approximately 2.44 times that of FSn in 2018, indicating the increasing role of artificial supplemental forage in supporting animal husbandry.

3.2. Dynamics of Forage Demand

The dynamics of forage demand are shown in Figure 9. From 2001 to 2018, the average forage demand was 350.35 × 10⁴ t, with the lowest value in 2001 (332.00 × 10⁴ t) and the highest value in 2017 (399.68 × 10⁴ t). The overall trend of total forage demand between 2001 and 2018 was upward, while there were different trends during the study period (Figure 9a). The forage demand increased between 2001 and 2007, with a rate of approximately 3.12 × 10⁴ t/a. Then, it decreased to a relatively low level during 2008–2013. After 2013, the forage demand increased rapidly, with a rate of approximately 15.12 × 10⁴ t/a, and it reached its highest level in 2017. From 2017 to 2018, the demand decreased. According to Figure 9b, the forage demand of the livestock at year end was approximately 3.9 times that of the slaughtered livestock, occupying approximately 80% of the total forage demand. The highest and lowest proportions of the forage demand of slaughtered livestock were in 2008 (24.14%) and 2003 (18.79%), respectively.

3.3. Forage Supply–Demand Relationships

In scenario I, herdsmen followed the transhumant grazing tradition of “living by water and grass”, grazing in different seasonal pastures throughout the year. They raised livestock mainly based on the forage provided by natural grasslands. The forage supply–demand relationships in scenario I are shown in Figure 10a. The average RSD_I value was 0.4717, which means that FSn accounted for only approximately half of the total forage demand. Temporally, the RSD_I decreased from 2001 to 2018 with fluctuating FSn and increased FD, indicating that the relationship between forage supply and demand was becoming tenser. However, the decreasing trend of RSD_I was not significant (p > 0.1). The highest RSD_I value was in 2013 (0.5999), the lowest was in 2008 (0.3891), and the highest RSD_I was approximately 1.5 times the lowest value, indicating that the growth conditions of the natural grasslands primarily affected traditional animal husbandry.

In scenario II, the RSD differed significantly from that in scenario I (Figure 10b). The overall forage supply–demand relationship was gradually optimized, and the forage–livestock balance was achieved in approximately 2013. The average RSD_II value of Altay Prefecture in 2001–2018 was 0.8289, indicating that the total forage supply reached more than 80% of the total demand. Compared with scenario I, the average RSD_II was approximately 1.76 times the average RSD_I. Temporally, the overall trend of the RSD_II was upward from 2001 to 2018, with the lowest and highest values in 2002 (0.6230) and 2018 (1.3799), respectively. The forage supply was less than the forage demand before 2012, with an average value of 0.7088 and the highest value of 0.8217. Since 2013, the forage supply began to exceed the forage demand, while in 2014 and 2015, the supply was slightly less than the demand (RSD_II values were between 0.9 and 1); after 2015, the RSD_II increased faster, with an average value of 1.1741. The forage that cannot be acquired in grazing prohibition areas can be effectively replaced by artificial forage with the development of artificial grasslands and the increase of supplemental forage.

4. Discussion

4.1. Influencing Factor Analysis

4.1.1. Natural Factors

Natural factors, such as climate and the physical and chemical properties of soil, and topographic conditions drive the forage supply–demand relationships mainly by influencing the FS. Precipitation and temperature could directly affect the growth status of grassland ecosystems, and then the FSn. Abundant precipitation during the growing season could provide better water conditions for grassland growth. However, the high temperature during growing seasons corresponded to high evapotranspiration, decreasing the amount of water available for grassland growth. Similar relationships between climate patterns and the FSn and grassland NPP were found in other regions in Northwest China [57,58]. For artificial grasslands, scarce precipitation would cause a shortage of irrigation water, resulting in reduced yield. Extreme climate events could also have apparent effects on FSn; for example, an extreme drought occurred in 2008 in Altay Prefecture, and it caused a marked decrease in FSn (Figure 5). Zhang et al. [44] also reported that the vegetation coverage of grasslands in Altay Prefecture had a notable decline in 2008 compared with other years. In addition, precipitation during the nongrowing season can affect the available FSn. For example, excessive precipitation (mainly solid water) during the nongrowing season will cause an increase in the snowfall depth, preventing livestock from eating forage. The floods in the spring and summer of the following year caused by the excessive precipitation in the nongrowing season will destroy the artificial grasslands along the rivers. Climate factors can also affect the forage demand in indirect ways. It can be found in Figure 2 that the number of slaughtered livestock had an obvious increase in 2007–2008 and a decrease in 2008–2009; meanwhile, the number of year-end livestock had an obvious decrease in 2007–2008. Correspondingly, the forage demand decreased from 2007 to 2009 (Figure 9a). This can also be explained by the extreme drought in 2008 mentioned above because the slaughter rate needed to be increased to reduce the forage demand and avoid losses when facing a decrease in forage supply.

4.1.2. Socioeconomic Factors

Socioeconomic factors include policies, institutions, markets, and human activities, which can affect both the supply and demand of forage and the supply–demand relationships. This study mainly focused on four policies and projects implemented after 2000, which have been widely discussed in the academic literature and have profoundly affected grassland restoration and pastoral livelihoods in the study area. The main concerns were how these measures could affect the supply or demand of forage and how the RSD changed correspondingly.

On the supply side, the RGLG and SISGC stipulated prohibiting grazing in certain areas and during specific periods. Grasslands can recover, and grassland biomass can increase after a reasonable period of grazing prohibition [59], indicating that herdsmen can benefit from a higher available forage in the long term despite the short-term decrease in available forage due to not grazing in these areas. Ecological construction, such as grassland seeding and improvement, has also contributed to grassland recovery from degradation [41]. These policies have also facilitated the construction of artificial grasslands; for example, with the implementation of the HSP, a 3.33-ha area of artificial grassland was reclaimed for every settled household. In addition, herdsmen would be paid ecological subsidies if they planted high-quality forage crops [60], which could improve the yield per unit area. The expansion of artificial grassland areas, the increase in yield per unit area, and the construction of irrigation infrastructure have improved artificial-grassland productivity and increased the supply of artificial forage. The FLB was implemented in 2005, interwoven with other policies, to continuously ensure that the number of livestock was within the carrying capacity of the grasslands [47]. In the context of these measures, the local grassland authority has set up several monitoring sites across the administrative region to regularly monitor grassland growth status, which would feed back to these measures. For example, when poor grassland growth was observed, livestock spatial distribution and mobility would be adjusted according to grassland productivity, and the herdsmen would be encouraged to increase the slaughter rate or buy more forage from other regions to decrease the forage demand and increase the forage supply.

It should be noted that the measures on livestock reduction did not mean reducing the total number of livestock but rather reducing the number of grazing livestock. Herdsmen can raise more livestock if they ensure that the number of grazing livestock does not exceed the grassland carrying capacity, as the number of livestock is closely related to their income level [61]. Therefore, driven by economic development, market incentives, and policy support, the livestock number increased, indicating an increasing demand for forage (Figure 9a). The FD would have largely exceeded the carrying capacity of natural grasslands if herdsmen relied entirely on the FSn; however, the increasing FSa has supplemented the FSn to support animal husbandry. As mentioned above, the construction of artificial grasslands and ecological subsidies enabled herdsmen to grow forage crops or purchase extra forage, which reduced the dependence of herdsmen on natural grasslands. The previous studies by Dai et al. [61,62] reported that the livelihoods in Altay Prefecture had undergone a transition, a large part of the traditional pastoral households have diversified their livelihoods, and the households with diversified livelihoods were less dependent on natural grasslands than were traditional pastoral households. Moreover, many pastoral households stopped grazing in winter pastures because some of their pastures were included in grazing prohibition areas and because of the harsh living conditions of winter pastures. On these bases, the proportion of grazing livestock declined, the proportion of stall-feeding livestock increased, and the overgrazing problem was gradually resolved.

4.2. Conceptualizing the Forage Supply–Demand Relationship from the SES Perspective

As stated in Section 4.1, the forage supply–demand relationships were driven by multiple factors, among which there were complex interactions. A systematic perspective is beneficial and indispensable when analyzing the pastoral socio-ecological system. Therefore, in this section, we employed the SESF to conceptualize the influencing factors to deepen the understanding of the forage supply–demand relationships in pastoral SESs.

The SESF contains four core subsystems: resource system, resource units, governance system, and actors [34,35]; the interactions between and within these four subsystems result in outcomes, which in turn affect the subsystems [34]. SESs also affect and are affected by external social, economic, and political settings and related ecosystems. The SESF is flexible and open, allowing modifications to the original framework to select the most relevant variables or adding more variables for analyzing the specific SESs of interest [63,64].

In Altay Prefecture, the resource system is the grasslands that support animal husbandry development; the governance system includes the rules, institutions, and policies related to grassland management and livelihood improvement; and the actors are the herdsmen or pastoral groups who utilize and rely on grassland resources for their livelihood [36]. Regarding the resource units, some have defined the forage as the resource unit and the livestock as the technologies available for the actors to harvest the forage, while some have treated both the forage and the livestock as the resource unit [36,38,65]. In this study, we defined that the resource units in Altay Prefecture contained forage from natural grasslands and artificial grasslands as well as livestock, following the study by Robinson et al. [36]. The subsystems and the interactions within the pastoral SESs are illustrated in Figure 11. Based on the conceptualization, the forage supply–demand relationship can be seen as the interactions among resource units [36]; its dynamics can be seen as the results of the dynamics of subsystems.

The FS mainly depends on the productivity of natural and artificial grasslands, and the livestock number mainly determines the FD. Therefore, the forage supply–demand relationships are directly driven by the dynamics of the resource system and actors, which interact through the provision and utilization of grassland resources. Regarding the resource system, the elements in the related ecosystem provide the growing environments for the natural and artificial grasslands in the resource system. Policies in the governance system can protect and restore degraded natural grasslands by implementing specific ecosystem management measures, such as grazing prohibition and grassland seeding. Some measures focus on improving the total yield of artificial grasslands. The combined effects of the related ecosystem and the governance system can affect the productivity of the resource system. Meanwhile, the dynamics of the resource system would, in turn, feed back to the governance system, leading to policy adjustments. Concerning the actors, the elements in the external social, economic, and political settings, such as economic development, can motivate them to improve their livelihoods by changing the breeding scale or engaging in diversified works. The governance system can set rules for actors concerning natural-resource utilization and support actors to improve their livelihoods. The combined effects of social, economic, and political settings and the governance system can affect the livestock number and change the actors’ dependence on natural or artificial grasslands. Similarly, the dynamics of actors would also feed back to the governance system and lead to policy adjustments.

4.3. Management Implications

Based on the quantification of the forage supply–demand relationship and the analyses of the influencing factors from the perspective of SES, this study provides three implications for grassland management, improving pastoral livelihoods, and promoting sustainability in Altay Prefecture. The first is ensuring adequate artificial-forage stock for livestock. This study finds that natural forage cannot fully support forage demand, and the role played by artificial grasslands is increasingly important. However, in the face of severe weather disasters, natural- and artificial-grassland yields decrease, which can result in a huge forage gap. The shortage of forage supply will threaten the livelihoods of herdsmen and the development of regional animal husbandry because they can choose to increase the slaughter rate to decrease forage demand to cope with the shortage, which would cause a decrease in the animal husbandry output of the coming years. Therefore, this study suggests that forage storage bases can be built, and forage reserves should be expanded in years with a high yield to cope with the potential forage shortage resulting from droughts or cold waves. The second is improving the efficiency of water use. The increase in the artificial-forage supply implies reclaiming more arable land and consuming more water for irrigation. However, the water resources in arid regions are distributed unevenly in time and space [66]. The water-resource utilization efficiency should be improved by continuously promoting water-saving irrigation technology and water resources should be allocated scientifically by building irrigation infrastructures. Meanwhile, it is necessary to consider the water demand in a basin’s upper, middle, and lower reaches and the water demand between different sectors, such as production and the environment, to avoid water conflicts. The third is maintaining the social–ecological relationship when implementing management measures by applying a holistic perspective. This study finds that the herdsmen, grasslands, and livestock are interrelated as are the factors that drive the dynamics of forage supply–demand relationships. Management concerning balancing the forage supply–demand relationship should consider all the key factors and their interactions. Moreover, management should focus on more than just a single social, economic, or ecological benefit and instead look at the comprehensive benefit.

4.4. Limitations and Prospects

Based on spatial and statistical data, this study explored the interannual variation in forage supply and demand in Altay Prefecture. However, some uncertainties and limitations need further consideration in future studies. The first is regarding the method and data used. The raw data accuracy and data processing affect the estimation accuracy of FSn and cause uncertainties. Additionally, due to the lack of data on the silage-corn yield in the early part of the study period, we estimated the yield based on the planting area in that year and the average yield per unit area in other years, which may also lead to uncertainties. Second, this study was carried out at the administrative scale to provide a macroscopic analysis of RSD. This study did not include the spatial mobility of livestock and the intra-annual dynamics of livestock numbers. Future studies could explicitly map the RSD at the pixel scale by considering the spatial mobility of livestock, which can provide a basis for the spatial allocation of livestock to manage the forage–livestock relationship more effectively. The variation in FSn in different seasons or months and livestock-number dynamics resulting from lambing and sale could also be considered, providing the basis for the dynamic adjustment of livestock scale and formulation of supplemental feeding plans. Third, this study provided a qualitative analysis of how the interactions of pastoral SES elements affected the forage supply–demand relationships, and a quantitative analysis is needed in future studies. Methods such as the system dynamics model could be applied in future studies to quantify the dynamics of forage supply–demand relationships under different natural and socioeconomic conditions.

5. Conclusions

This study explored the dynamics of forage supply, demand, and supply–demand relationships in Altay Prefecture, a typical pastoral SES, and analyzed the influencing factors. Two scenarios were set to study the forage supply–demand relationships by considering the different combinations of natural forage in free-grazing areas, cross-region grazing areas, grazing prohibition areas, and artificial forage. The SESF was employed to conceptualize the influencing factors as interacting SES elements. The main conclusions are as follows: (1) From 2001 to 2018, the overall trend of FSn fluctuated, and the FSa and FD trended upward; the proportion of FSn in FS decreased, while the proportion of FSa increased. Without FSa supplementation, the average RSD was 0.4717 during 2001–2018; however, the value increased to 0.8289 when FSa was supplemented. (2) The quest for higher income and better lives has increased livestock numbers and forage demand, and the yearly fluctuant productivity of natural grasslands can hardly support animal husbandry if there is no abundant artificial supplemental forage. Under the implementation of relevant policies and projects, the increasing artificial supplemental forage has promoted a balance between forage supply and demand since approximately 2013, indicating an optimized grassland utilization pattern. (3) The forage supply–demand relationships can be affected by different factors and system governance is needed in pastoral SESs. Management regarding protecting and restoring grasslands, improving pastoral livelihoods, and promoting the sustainability of pastoral SESs should adopt a holistic perspective to consider the pastoral SES elements in ecological and social aspects and their interactions.