It is generally recognized that, by providing access to vegetation with adequate forage quality and quantity, traditional mobile pastoralism has adapted well to arid and semi-arid rangeland ecosystems with spatially and temporally stochastic rainfall regimes. However, the pastoralists' strategies for spatial movement have been changing in response to external forces such as changes in political and economic regimes. In the steppe region of Mongolia, the shift toward a market economy since the early 1990s has dissolved the system of negdel, the cooperative agricultural and stock-raising associations under the socialist regime. Because this dissolution has the potential to cause overstocking and concentration of livestock in particular areas, increased risk of grazing-induced pasture degradation has become a major concern in rangeland management sectors.

For the last two decades, rangeland ecologists have been discussing the relative importance of biotic and abiotic factors in controlling vegetation dynamics. According to non-equilibrium concept, which emerged as a new paradigm to describe ecosystems in highly variable and poorly predictable environments, vegetation dynamics are driven primarily by abiotic factors such as rainfall, rather than by internal biotic regulation (e.g. grazing impact). Although a number of studies have emphasized the non-equilibrium nature of most rangelands, they have produced inconsistent results and there is still no consensus on the relative importance of grazing impact. Alternatively, recent studies have suggested that a continuum of systems exists, rather than a stark dichotomy between equilibrium and non-equilibrium rangelands. Moreover, theoretical studies have predicted that, in addition to climatic variability, resource variations occurring in space and time as a result of landscape heterogeneity should be taken into account in assessments of grazing impact. However, there have been few empirical studies of the role of resource variations in either mitigating or enhancing the impacts of grazing on vegetation.

This thesis explores the impact of grazing strategies on vegetation in Mongolian rangelands under high climatic variability. Specifically, I focused on key resource areas, defined in terms of the key factor determining livestock population, and thus the ability to forage during drought. Based on the prediction proposed by Illius and O'connor (1999) that animal numbers are regulated in a density-dependent manner by the limited forage availability in key resource areas, I established the hypothesis that grazing impacts would be greater in key resource areas than in other areas even in Mongolian rangelands. The study had three aims: (1) by using vegetation survey and satellite image analysis, to quantify the spatial and temporal arrangement of resources represented by vegetation (Chapter 2); (2) to clarify the pastoralists' strategies for the proper use of varying resources (Chapter 3); and (3) to examine the relationship between these strategies and vegetation degradation (Chapter 4). Finally, I discuss the implications of the results of these empirical studies for appropriate rangeland management schemes under high levels of climatic variability (Chapter 5).

The study area was located in the Saintsagaan sum (district), which includes the city of Mandalgobi (45°46'N, 106°16'E) in Mongolia's Dundgobi province. Mandalgobi is located in the desert-steppe ecological zone. According to the prediction that non-equilibrium dynamics predominate in areas where mean annual precipitation is less than 250 mm and precipitation coefficient of variation exceeds 33%, the study site was considered to be in a highly variable environment.

I quantified the spatial and temporal distribution of the vegetation by combining the vegetation survey and satellite image analysis. First, I established 246 quadrats (each 20 × 20 m) within the study site (20 × 20 km) and recorded the percentage cover of all plant species in each quadrat. The vegertation data were subjected to detrended correspondence analysis and then classified by using a cluster analysis into the following four community types that differed in species composition and abundance: (1) Caragana community, dominated by the legume shrubs Caragana microphylla, perennial grass Cleistogenes squarrosa, and perennial forb Allium mongolicum; (2) Allium community, dominated by the perennial forb Allium polyrrhizum; (3) Achnatherum community, dominated by the large, dense bushy grass Achnatherum splendens and the annual forb Salsola collina; and (4) Reaumuria community, dominated by the halophytic semi-shrubs Reaumuria soongorica, Salsola passerina, and Kalidium foliatum.

Next, I analyzed the spatial distribution of the four community types and their temporal changes by using satellite images of two periods (a drought summer with low precipitation in 2009 and a normal summer with average precipitation in 2010) and thematic maps (e.g. geomorphology map). The Achnatherum and Caragana communities, respectively, were clearly extracted as having the highest and second-highest normalized difference vegetation index (NDVI) values. Reaumuria and Allium communities, both of which had low NDVI values, were differentiated by using texture information. The Allium community covered the largest area, while Achnatherum and Reaumuria the smallest, among the four communities. The four communities (Caragana, Allium, Achnatherum and Reaumuria) were spatially arranged along the sequence of land catena, that is, on denudation plain, hillslope, pediment, and depression, respectively. Biomass, as estimated from the NDVI values, was significantly lower in the drought summer than in the normal summer. However, the Achnatherum community maintained a relatively high NDVI, even during drought, compared with those of the other communities. These results suggested that both topographic heterogeneity and climatic variability contribute to providing resource variations in space and time, and that among varying resources, Achnatherum community has a function of key resource area during droughts.

The pastoralists' patterns of use of heterogeneous vegetation resources were clarified by using interview surveys. First, I interviewed four leaders of bags (villages) as key informants to clarify the pastoralists' resource selection during drought and normal summers. The result confirmed that pastoralists used different resources in normal and drought summers: during normal summers all types of community were used, but during drought the main community used was the Achnatherum community.

Second, to determine the movement patterns at the study site, I interviewed 77 pastoralists who stayed in the Saintsagaan sum. I asked pastoralists the numbers of livestock they had, the places they visited during drought (2009) and normal (2010) summers, and the basis of their pasture selection during droughts. I used a cluster analysis of interview data to divide the the pastoralists' movement patterns into the following two broad groups: (1) mobile wealthy pastoralists, tending to have larger flocks and move longer distances; and (2) less-mobile poor pastoralists, having smaller flocks and moved shorter distances. While the former groups could move to remote pasture dominated by perennial grasses with high palatability and quick response to rainfall, the latter groups were compelled to stay in home pasture and to utilize Achnatherum community with low palatability but high tolerance to dry conditions which retains high levels of vegetation cover even during drought. These results indicated that pastoralists used the resource variations occurring in space and time, and that the lack of uniformity in the pastoralists' movement patterns would also have affected the spatial distribution of the impact of grazing on vegetation. Those also suggested that the risk of grazing-induced vegetation degradation would increase, especially in the Achnatherum community with narrow distribution ranges along the bottoms of river valleys, where less-mobile poor pastoralists concentrate during drought to reduce the cost of mobility.

To examine whether the grazing strategies were mitigating or enhancing the impact of grazing on vegetation, I performed a vegetation and soil survey. I set up 20 quadrats (each 1 × 1 m) each within a grazed and an ungrazed plot in each community (Caragana, Allium, Achnatherum and Reaumuria). I identified all species present in the quadrats, estimated their percentage cover and calculated the dissimilarity indices between the ungrazed and grazed plot in each community. I then compared the dissimilarity indices among the four communities. The dissimilarity index of the Achnatherum community was the highest, indicating that species composition was significantly more affected by grazing in the Achnatherum community than in the other communities. The number of dung pellets was also highest in the Achnatherum community, suggesting that this community was subject to the highest grazing intensity. These results suggested that the distribution of grazing impact was uneven across communities, and that the Achnatherum community was affected more by grazing than were the other communities, supporting the predictions of theoretical studies that equilibrial forces exist over a limited part (i.e. key resource area) of the non-equilibrium environment.

The key finding of this thesis is that grazing-induced degradation, that is, the equilibrium nature is detectable over key resource areas even in non-equilibrium environments by taking into account resource variations at a landscape scale in relation to the scale of local rangeland use. Therefore, to develop effective rangeland management systems the debate needs to be shifted from the equilibrium versus non-equilibrium dichotomy toward a greater awareness of resource variation. I showed that the impact of grazing on the Achnatherum community was high; because this community occupies a small area, its degradation could reduce livestock populations by diminishing the ability of the ecosystem to provide spatial buffers during climatic disturbance. These findings suggest that if rangeland management is focused only on maintaining resource accessibility, then grazing-induced degradation of resources can occur. Therefore, controlling livestock numbers remains indispensable to rangeland management, even in highly variable environments.

Variations in grazing strategies have generally not been concerned in assessments of degradation. However, I showed that such variations were tightly related to the spatial pattern of grazing-induced degradation. This finding suggests that rangeland managers should take into account variations in grazing strategies depending on pastoralists' economic situations, because such variations should need different countermeasures for sustainable rangeland use.

In conclusion, this thesis clearly shows that grazing strategies and resource variations interact tightly under high climatic variability. This empirical evidence offers a major first step toward sustainable rangeland management in semi-arid regions.

本研究では気候変動性が高い乾燥地の放牧地生態系を対象に、植生資源の空間的異質性と牧民の放牧戦略の異質性の関係をスケールに着目して解析することで、放牧システム動態に関するモデルの統合およびそれに基づく管理手法の提案を試みた。具体的には、干ばつ時に利用される資源(key resource)では放牧圧の影響が強く表れるというIllius and O'Connor (1999)のモデルを、放牧戦略の異質性を考慮して検証した。この目的を達成するため、以下の手順によって研究を実施した。(1)植生の時空間的異質性を現地での植生調査および衛星画像解析によって評価した。(2)放牧戦略の異質性を明らかにするため、現地牧民への聞き取り調査を実施した。(3)このような資源の空間的異質性と放牧戦略の異質性の関係が土地荒廃へ与える影響を、現地植生調査によって明らかにした。

まず、植生資源の時空間的異質性を定量的に評価するために、植生調査と衛星画像解析を実施した。その結果、調査地の植生は大きく4つのグループに分けられ、それぞれ (1)マメ科低木である Caragana microphyllaを指標種とするCaragana群落、(2)多年生双子葉草本 Allium polyrrhizumを指標種とするAllium群落、(3)多年生イネ科叢生草本の Achnatherum splendens を指標種とするAchnatherum 群落、(4)塩性低木のReaumuria soongoricaを指標種とするReaumuria 群落となった。各群落は空間的に不均一に分布していており、Allium群落が最も広く全体の70.8%を占め、Caragana, Achnatherum, Reaumuria群落はそれぞれ14.0%、 6.4%、8.8%を占めていた。地形との関連を解析した結果、Caragana群落は小丘、Allium群落は傾斜面、Reaumuria群落は低地と対応し分布していた。さらに、干ばつ時にAchnatherum群落は他の群落に比べ比較的NDVI値が高い傾向があった。以上の結果から、key resourceの特徴、すなわち、面積が比較的小さく、干ばつ時に利用可能な資源という特徴を最もよく示しているのはAchnatherum群落、ついで Caragana群落、Reaumuria群落の順であると考えられた。

つぎに、牧民の干ばつ時の放牧戦略を明らかにするために、経済性、干ばつ時の移動性、干ばつ時の資源利用について現地牧民74名を対象に聞き取り調査を実施した。その結果、牧民の干ばつ時の放牧戦略は、所有家畜頭数の違いによって2つの異なる傾向に分けられることが分かった。すなわち、家畜頭数の多い大規模牧民は干ばつ時に雨が降った場所へ長距離移動すると回答する傾向があり、家畜頭数の少ない小規模牧民は干ばつ時にAchnatherum群落を利用して移動は短距離であると回答する傾向があった。このように、経済性に由来する移動性の違いによって、牧民は干ばつ時に異なるスケールの空間的異質性を利用し生存を図ろうとしてした。とくに、key resourceの特徴を持つAchnatherum群落は主に小規模牧民によって利用されていることが明らかになった。

放牧圧が各群落へ与える影響を明らかにするために、各群落に禁牧柵を設置して、禁牧区と放牧区にて植生調査を実施した。その結果、Reaumuria群落における放牧区と禁牧区の種構成の非類似度指数が最も高く、次にAchnatherum群落が高かった。Reaumuria群落における放牧区と禁牧区の種の違いは、同じ機能タイプ群である塩性低木種の入れ替わりであった。一方でAchnatherum群落は、放牧区において優占種の多年生イネ科草本の減少、および過放牧の指標として用いられるChenopodium albumの増加が確認され、放牧に関連した植生の応答が観察された。 さらに、各群落にかかる放牧圧を推定するために、各群落におけるフンの数を計測した結果、Achnatherum群落ではフンの数が有意に多かった。このことから、資源の空間的異質性を通して放牧圧は不均一に分布しており、干ばつ時に利用されるAchnatherum群落では、長期的な強い放牧圧の影響により植生の種構成が大きく変化している可能性が示唆された。以上から、key resourceでは放牧圧の影響が強くあらわれるというIllius and O'Connor (1999) の予測を支持する結果が得られ、変動性の高い非平衡環境下においても、平衡環境がパッチ状に形成され、そこでは放牧圧のコントロールが必要であることが示された。とりわけkey resourceであるAchnatherum群落は移動性の低い小規模牧民によって利用されていることから、key resourceの保全は零細牧民のセーフティネット確保という観点からも重要であることが示唆された。

本研究は、気候変動性の高い乾燥地の放牧地生態系において、地形由来の空間的異質性に基づく局所スケールでの資源利用および、降雨由来の空間的異質性に基づく広域スケールでの資源利用が同時に生じており、とくに前者の資源利用において家畜頭数の密度依存的な変化と植生の劣化が生じうることを実証的に示したものである。本研究成果は、スケールの階層性に着目することにより、放牧地生態学における平衡-非平衡モデルの統一的理解が可能であることを実証的に示した点において高く評価できる。さらに、局所スケールにおけるkey resourceの放牧圧管理の有効性を示すなど、持続的放牧地管理に向けた応用的意義も高い。よって、審査委員一同は、本論文が博士(農学)の学位論文に十分値する論文であると判断した。