Previous PageTable Of ContentsNext Page

Grass and legume variety trials in eastern Tibet

W.E. Limbach,1 L. Yingchun2 and M. Yushou3

1. Qinghai Livestock Development Project* (QLDP), RDP Livestock Services, P.O. Box 523, 3700 AM, Zeist, The Netherlands
2. Project Management Unit, Bureau of Animal Husbandry, Department of Foreign Trade and Economic Co-operation, 4 Jiaotong Road, Xining 810008, Qinghai, P.R. China
3. Institute of Grassland Science, Qinghai Academy of Animal and Veterinary Sciences, Xining 810003, Qinghai, P.R. China

Summary

Approximately 28% (42.5 million hectare (ha)) of Qinghai's high elevation rangelands are considered to be in deteriorated condition. Re-vegetation of these degraded rangelands requires improved forage grasses adapted to the harsh environment of the Qinghai-Tibetan Plateau. In May 1999, high altitude grass varieties of Poa, Festuca, Deschampsia, Bromus, Stipa, Agrostis, Dactylis, Phleum, Elymus, and Agropyron and legume varieties of Medicago, Onobrychis, Astragalus, and Lupinus were planted at two replicate locations in Guoluo Prefecture: Dawu Seed Farm (3512 metres above sea level (masl)), Maqin County and Jimei Research Area (3968 masl). Each location contained 75 plots (3 m by 3 m) with varieties randomised within 3 blocks, 25 plots to a block in a randomised block design. Seedlings were monitored for establishment (year 1) and survival over winter (year 2). Seeds were also analysed in the laboratory for germination characteristics. There were significant differences (P<0.0001) among varieties for seedling establishment and over-winter survival. Differences between locations were significant (P<0.0001). Coefficients of determination (R2) ranged between R2 = 0.78 and 0.80. Eight grass varieties survived over winter at optimal levels, in descending order: Festuca ovina, Deschampsia caespitosa, Tibetan Mixture, Festuca rubra var. rubra, Poa pratensis var. alpinum, Bromus inermus, Poa compressa and Poa pratensis. Laboratory germination tests did not accurately predict success in the field. Differences in performance within species points out that different genotypes (seedlots) of the same species can perform very differently. Implications for re-vegetation of degraded sites in Guoluo are discussed.

Keywords: Forages, Qinghai, rangeland, re-vegetation, sheep, subalpine, yak

Introduction

Approximately 28% (42.5 million ha) of Qinghai's high elevation rangelands are considered to be in deteriorated condition (Lang et al. no date). Erosion features known as 'black beach' or 'black soils' are common, many of which cover 20 to 30 ha and more. Black beach forms when the native Kobresia sod is broken and wind, water and the freeze-melt cycle induces the loose sod to erode down slopes, thus exposing the dark, black soil beneath. Pikas (Ochotona spp.), small burrowing lagomorphs, further complicate the situation by competing with livestock for the native rangeland forages. Furthermore, livestock production in Qinghai is hampered by the lack of adequate forages in both quantity (maintenance rations) and quality (production and reproduction rations). In order to remedy the condition of these rangelands and improve the nutrition of yak and sheep, adapted forage grasses are needed for re-vegetation (Limbach 1998; Sheehy 1998).

Townships involved with the Qinghai Livestock Development Project in Guoluo Prefecture extend from about 3000 to over 6000 masl. These rangelands constitute a valuable and albeit impoverished forage base for yak, sheep and horses, and the subsistence livelihood of the Tibetan herders. Weight loss and mortality among livestock are common occurrences in most years, especially in late winter and early spring when livestock condition is at its lowest, the forages in the winter pastures are overused, and heavy snowfalls, known as 'snow disasters', cause high livestock mortality. Re-vegetation of these impoverished rangelands will greatly increase the livestock production, improve the grazing during winter, help reduce the disastrous effects of heavy snows, and aid in improving the livelihoods of Tibetan herders in the project area. Improved forage grasses are required for the re-vegetation and rehabilitation of these degraded rangelands to improve livestock performance and to renew and re-stabilise degraded sites.

This field trial was implemented to test the adaptation and suitability of 25 improved varieties of perennial forages for rangeland re-vegetation in Guoluo.

Materials and methods

In 1999, 19 improved varieties of perennial grasses and legumes were imported from Pickseed Canada, Vancouver, British Columbia (Table 1). Six other varieties, two grasses, three legumes, and Tibetan Mixture were previously imported from Germany in 1996 (Archer 1996). In late May and early June, these varieties were planted at field sites near Dawu, Maqin County and Jimei, Dari County. In July 1999, these varieties were also tested for germination characteristics in the laboratory of the Qinghai Academy of Animal and Veterinary Sciences (QAAVS).

Species screening: emergence, establishment and over-winter survival

A randomised block design experiment was set up at Dawu (ca. 3659 masl) and Jimei (ca. 3963 masl). At each location, an area of about 50 m by 20 m was ploughed, cultivated, and 75 plots (3 m by 3 m) were arranged in 3 blocks of 25 each; blocks = replicates (n = 3). Each block consisted of five rows of five, 3 3 plots. All interspaces between rows and blocks were 1 m wide. Plots within rows were adjacent to each other and had no interspaces between them. Varieties were planted into these plots in a completely randomised arrangement within block. Due to different sized seedlots and large differences in seed sizes, different seeding rates were used (Table 1). Seedling densities for emergence (early June 1999), establishment (first week of September 1999), and over-winter survival (middle May 2000) were sampled using a 0.0625 m2 quadrat (25 cm   25 cm), one randomly placed quadrat per plot, excluding 0.5 m around the edge of each plot. Results (seedling densities, individuals per quadrat) were assessed using a 2-way analysis of variance (ANOVA). Since survival over-winter is the most important characteristic of a variety once it has established in the first year, only over-winter survival is discussed in this paper.

Table 1. Grass and legume species names, variety number, seeding rate and source of seeds.
 No. Grasses Species names G plot1 Source

1

Poa alpina var. mantelsaatgut

18

Germany

2

Poa compressa

18

Canada

3

Poa pratensis var. alpina

18

Canada

4

Poa pratensis

18

Canada

5

Stipa viridula

36

Canada

6

Koeleria cristata

18

Canada

7

Dactylis glomerata

36

Canada

8

Bromus inermus

36

Germany

9

Festuca ovina

36

Canada

10

Festuca rubra var. rubra

36

Canada

11

Agrostis alba

20

Canada

12

Phleum pratense

18

Germany

13

Phleum pratense

18

Canada

14

Deschampsia caespitose

18

Canada

15

Deschampsia caespitose

18

Germany

16

Trisetum flavescens

27

Canada

17

Agropyron smithii

36

Canada

18

Agropyron dasystachyum

36

Canada

19

Elymus trachycaulus

36

Canada

20

Tibetan Mixture1

18

Germany

Legumes

1

Medicago sativa var. Anik

18

Germany

2

Onobrychis sativa var. Remont

36

Canada

3

Astragalus cicer

18

Canada

4

Medicago sativa var. Able

18

Canada

5

Lupinus luteus

36

Germany

1. Tibetan Mixture comprised 11grasses: Agrostis canina, A. gigantea, A. capilaris, Dactylis glomerata, Deschampsia flexuosa, Festuca rubra var. rubra, F. rubra var. communtata, Poa alpina, Phleum pratense, Poa annua, Trisetum flavescens and 5 forbs: Achillea millefolium, Anthyllis vulneria, Lotus cornicullatus, Trifolium hybridum and T. repens.

Laboratory germination

Three replicates of 100 seeds of each variety were sown into flats filled with sterilised, washed sand and kept well watered during the course of the investigation. Seeds were incubated in a dark growth chamber at 20°C for 28 days. Counts of emergent seedlings were made daily. Maximum percentage of germination (Gmax) was assessed.

Statistical analyses were performed using SPSS 8.0 for Windows, Chicago, IL, USA. Correlation was used to assess the association among laboratory germination and seedling performance in the field. Mean comparisons were determined using the Least Significant Difference test, ά = 0.05.

Results and discussion

The 1999 summer-growing-season was a good period for seedling emergence and establishment. Rainfall was well above the long-term average, which ranges between 500 and 600 mm, in the project area (Buda, Veterinary Station, Dari County, personal communication). All varieties were able to fully express their establishment vigour.

Grass seedling densities (number of seedlings/quadrat) for establishment in 1999 were high at both locations: mean = 69.0 each at Dawu, 57.7 each at Jimei. Legume seedling densities were lower than grasses; mean = 18.7 each at Dawu, 17.3 each at Jimei, though the two Medicago varieties appeared to be most successful legumes with means of about 36 each at Dawu and 20 each at Jimei. Lupinus luteus was the first of the legumes to emerge but hares heavily grazed it. This caused all Lupinus seedlings to die out before the end of the growing-season.

In 2000, there were significant differences in over-winter survival amongst varieties (P<0.0001) and between sites (P<0.0001); the location by variety interaction term was also significant (Table 2). Differences between locations probably were due to differences in rainfall and soil fertility; these differences, however, are not biologically significant. Significant differences among varieties would be expected in a field test of so many species and varieties. The coefficient of determination (R2) shows that the ANOVA model fits the data well and explained about 80% of the variation; only 20% of the variation was attributable to unexplained sources.

Table 2. Response of over-winter seedling survival (number/quadrat) to variations in location and variety.

Source

Df

SSIII

MS

F

P>F

Location

1

74.907

74.907

31.125

<0.0001

Variety

24

661.333

27.556

11.450

<0.0001

Location Variety

24

196.427

8.184

3.401

<0.0001

Error

100

240.667

2.407

   
Model R2 = 0.795

Over-winter seedling survival densities are presented in Table 3. Eight grass varieties survived at optimal levels: Festuca ovina (variety 9) at 112 individuals/m2, Deschampsia caespitosa (variety 14) at 96 individuals/m2, Tibetan Mixture (variety 20) at 88 individuals/m2, Festuca rubra var. rubra (variety 10) at 80 individuals/m2, Poa pratensis var. alpinum (variety 03) at 72 individuals/m2, Bromus inermus (variety 08) at 72 individuals/m2, Poa compressa (variety 02) at 64 individuals/m2, and Poa pratensis (variety 04) at 56 individuals/m2 (Table 3).

Table 3. Seedling survival densities (individuals/m2) in May 2000 at Dawu, Maqin County and Jimei, Dari County, Guoluo Prefecture, Qinghai, P.R. China.

Varieties

Dawu

Jimei

Variety

Dawu

Jimei

9

4

10

15

2

2

14

6

6

5

1

2

20

2

9

12

2

1

10

3

7

17

2

1

3

3

6

7

1

1

8

2

7

16

1

0

2

2

6

1

0

0

4

3

4

21

0

0

6

1

5

22

0

0

13

2

3

23

0

0

19

2

3

24

0

0

18

2

3

25

0

0

11

3

1

     

These species-varieties are apparently well adapted to establishment and survival in the severe high altitude environment of Guoluo Prefecture. These characteristics make them prime candidates for rangeland re-vegetation and improvement projects in Guoluo and the Qinghai-Tibetan Plateau. It is not surprising that these same species are distributed worldwide at high altitudes on all the major continents of the Northern and Southern Hemispheres (Wu and Wang 1999). It would be interesting to compare these imported varieties with local seed sources of the same species.

The results obtained with the Tibetan Mixture should be qualified. It was difficult to tell which individuals were from the mixture and which were volunteers from the local seed bank. In addition, no legumes emerged from the Tibetan Mixture.

Survival of even the most robust varieties was low, compared to the numbers of seedlings that had established at the end of the 1999 growing-season (Limbach 2000). This may be due to a number of factors:

  1. Livestock trespassed on both experimental areas between October 1999 and May 2000. There was evidence of yak trespass at Jimei and sheep trespass at Dawu. Therefore, grazing by livestock may have killed large numbers of the grass and legume seedlings that had established in 1999 by pulling them out of the ground.
  2. Desiccation of seedlings over the 1999/2000 winter, which was drier than normal (Buda, personal communication), may have caused a large number of seedlings to die.
  3. As a consequence of a below-normal, dry winter, the soils of the field sites were not afforded the ameliorating effect of a cover of snow. This may have exacerbated seedling die-off by frost action pushing the weakly rooted seedlings out of the soil. Furthermore, the valley of Dawu is much more open and broad than the valley in Jimei. This may account for the poorer survival at Dawu (Table 3) even though Dawu had better seedling establishment than Jimei in the autumn of 1999 (Limbach 2000). The soils at Dawu did appear to be looser, fluffier, and less compact than the soils at Jimei.

The varieties are ranked in descending order in Table 4. Low rankings refer to good performance (high seedling densities or high Gmax) while high rankings refer to poor performance (low seedling densities or low Gmax). Correlation coefficients (R) show unpredictable relationships between survival and establishment (R = 0.47; P = 0.018) and survival and maximum germination percentage, Gmax (R = 0.10; P = 0.628). These characteristics, obtained in the greenhouse and laboratory, did not predict over-winter success in the field.

Table 4. Rankings of varieties according to maximum germination (laboratory), field establishment (September 1999), and over-winter survival (June 2000).

Varieties

Scientific Name

Gmax

September 1999

Survival 2000

9

Festuca ovina

11

8

1

14

Deschampsia caespitosa

18

4

2

20

Tibetan Mixture

9

9

3

10

Festuca rubra var. rubra

17

14

4

3

Poa pratensis var. alpina

14

6

5

8

Bromus inermus

8

19

6

2

Poa compressa

20

1

7

4

Poa pratensis

23.51

22

8

6

Koeleria cristata

2

12

9

13

Phleum pratense

6

7

10

19

Elymus trachycaulus

25

10

11

18

Agropyron dasystachyum

3

13

12

11

Agrostis alba

19

2

13

15

Deschampsia caespitosa

16

16

14

5

Stipa viridula

22

17

15

12

Phleum pratense

13

11

16

17

Agropyron smithii

21

20

17

7

Dactylis glomerata

12

5

18

16

Trisetum flavescens

5

3

19

1

Poa alpina var. mantelsaatgut

23.51

25

20

21

Medicago sativa var. Anik

10

15

21

22

Onobrychis sativa var. Remont

7

23

22

23

Astragalus cicer

15

21

23

24

Medicago sativa var. Able

4

18

24

25

Lupinus luteus

1

24

25

1. Ties in ranking within columns were averaged.

Testing different varieties within species

Different varieties (genotypes) within species and different seedlots within species may perform very differently. When a sample of particular species fails in field tests, it does not mean that all varieties of that species will fail. This fact is well demonstrated in the present study by the performance of varieties numbers 14 and 15, different genotypes of Deschampsia caespitosa (Table 5). Screening plant materials at the species level is severe as a selection criterion and will greatly limit the amount of plant materials for future selection.

Table 5. Mean over-winter seedling density (number/ quadrat) of two varieties of  Deschampsia caespitose.

Variety

Dawu

Jimei

DECA 14

5.67

6.33

DECA 15

1.67

2.00

Differences are significant (P<0.001) using Chi-Square Goodness of Fit.

A caveat on rangeland re-vegetation

The next step is to extend these results to reseeding of degraded range sites. Restoring the productive capacity of degraded ranges, however, is a major undertaking anywhere it is contemplated. It is expensive in terms of time, money and manpower. And it is very risky. This latter point cannot be over-emphasised. When the plant community has been degraded or transformed to a lower ecological-successional status, an ecological threshold has been crossed. This ecological threshold is a theoretical demarcation between plant communities of different successional status, a more desirable plant community and a less desirable one. In the degraded state, the plant community is one of lower ecological status and poorer in terms of grazing. Once the rangeland has changed to a lower, less productive status, however, it is very difficult to return it back to a more productive status. Direct manipulations, like re-vegetation with modern equipment and rehabilitation techniques, are required to restore a productive plant community. But what happens if after all the effort the seeding fails due to the vagaries of the climate and environment?

At present, a typical reseeding operation as performed in Guoluo by the QAAVS using traditional agronomic techniques may consist of four or more separate operations, each one requiring tractor and implement to manipulate the entire operational area: initially ploughing, then harrowing, followed by broadcast seeding and application of fertiliser, then packing or tamping the soil, and a second fertiliser application in another month or so. The cost of such a reseeding operation is about 120150 RMB Yuan/mu (US$ 1 = 8.2 Yuan during this study; 1 hectare = 15 mu) (Ma Yushou, personal communication). This is approximately twice the cost of similar operations in countries such as Australia and the United States. The cost effectiveness of reseeding operations could be considerably improved by utilisation of an inter-seeding, rangeland drill. Reseeding with a rangeland drill can be effected in a single pass, thus greatly reducing the costs in fuel and manpower.

Acknowledgments

The authors wish to thank the Beijing delegation of the European Union, the Qinghai Livestock Development Project, the Department of Foreign Trade and Economic Co-operation (Qinghai Province), the Bureau of Animal Husbandry, and the Qinghai Academy of Animal and Veterinary Sciences for their support in this research. Thanks also go to Li Faji and personnel of the Maqin Grassland Station as well as Buda and personnel of the Dari Veterinary Station and Quning and personnel of the Dari Grassland Station. Without their help in the field, this research could not have been completed.

References

Archer A.C. 1996. Rangeland Extension Specialist, Progress Report. Qinghai Livestock Development Project, ALA/CHN/9344. Bureau of Animal Husbandry, Qinghai Academy of Animal and Veterinary Sciences. Xining, Qinghai, P.R. China. 63 pp.

Lang Baining, Wang Qiji and Ma Yushou. (no date). Review of the study on black-soil type deteriorated pasture. Internal Memo, Qinghai Academy of Animal and Veterinary Sciences. Xining, Qinghai, P.R. China. 5 pp.

Limbach W.E. 1998. End of Mission Report, Rangeland Agronomy. Qinghai Livestock Development Project, ALA/CHN/9344. Bureau of Animal Husbandry, Qinghai Academy of Animal and Veterinary Sciences. Xining, Qinghai, P.R. China. 24 pp.

Limbach W.E. 2000. End of Mission Report, Rangeland Agronomy. Qinghai Livestock Development Project, ALA/CHN/9344. Bureau of Animal Husbandry, Qinghai Academy of Animal and Veterinary Sciences. Xining, Qinghai, P.R. China. 31 pp.

Sheehy D.P. 1998. Report, Rangeland Management Specialist, ALA/CHN/9344. Qinghai Livestock Development Project. Bureau of Animal Husbandry, Qinghai Academy of Animal and Veterinary Sciences. Xining, Qinghai, P.R. China. 20 pp.

Wu Yuhu and Wang Qiji. 1999. The grasses of the Karakorum and Kunlun mountains. Qinghai People's Publishing House, Xining, Qinghai, P.R. China. 351 pp. [in Chinese].

Previous PageTop Of PageNext Page