6. Conclusions
6.1. Direct results of this work
Our study has identified some of the indigenous rhizobia and media to verify them, the distribution of wild legumes and shown that improvement potential for the vetch-vetch rhizobia symbiosis seemed to be limited. It was also shown that N does not appear to be a limiting factor on riverbank soils.
6.2. Some sober thoughts on nitrogen cycling in cold environments
Under constant environmental conditions, every ecosystem adjusts to an equilibrium status and despite some dire predictions on the fate of the Qinghai - Tibet plateau under global warming (see section 1.2) as well as reports on overgrazing exist (see section 2.3.), it might be legitimate to assume that present nitrogen cycles are not very far from this equilibrium status. As far N export is concerned, denitrification (Ruoergai soils see #Long Zhangfu 1994) and cattle raising are the major causes for losses. On the input side the rather unpolluted atmosphere suggests that nitrogen inputs from precipitation are probably not dominant in most habitats (bare rocks of course being an exception). So most of the balance is probably made up by nitrogen-fixing bacteria. Before we discuss the question who these bacteria might be, keep in mind that these tundra soils have a very high nitrogen content (about 0.5%), so that the soil nitrogen pool exceeds annual fresh fixation by several magnitudes (#Nadelhoffer 1991). Thus soil-temperature related destruent activity fluctuations will complicate accurate input/output measurements.
In chapter 2 we concluded that legume distribution is primarily related to soil water content; and water content is probably also very important for the prevalent form of nitrogen fixation in each habitat. Remember that in the wet marshlands no legumes were found; this means that their nitrogen either comes with inflowing water from other environments (possibly in the form of amino acids, which can be directly absorbed by some arctic sedge communities (#Chapin 1993) - note that the moist swamp habitats are dominated by sedges (#Chai 1965,#Yang 1993B,#Zhao 1996)) or, from other non-symbiotic forms of N fixation, potential candidates being the N-fixing bacteria in the nutrient-rich rhizosphere (#Yan 1993)) of all arctic graminoids (#Nosko 1994)) or cyanobacteria in moist sedge communities (#Henry and Svoboda 1986).
Cyanobacteria-moss associations (which in Xiaman might be prevalent on legume-poor moist northern slopes) are also known to be active nitrogen fixers in tundra environments (#Alexander 1978,#Carr 1980 and #Henriksson 1987).
On dryer slopes or the dry riverbank plain, we found legumes to be more numerous, but also in these environments legume-nitrogen fixation probably has to compete with grass/sedge-associated rhizosphere nitrogen fixation (note that #Long (1994) determined Azotobacter to be the second-most dominant bacterial genus is Ruoergai soils), which in quantity might be surpassed by ectomycorrhizal N release related e.g. with Polygonum viviparum (#Michelsen 1996), a kind of knotgrass also found at Xiaman.
Briefly, we want to emphasize that although leguminous N-fixation is so conspicuous, it is not necessarily the most important form for cold felt soil tundra environments (#Alexander 1978), and our legume biomass, over the year, probably accounts for only 1-2% of total biomass, and a nodule weight of perhaps less than 1mg/1g plant weight would translate to some meager 50-60 grams of nodules per hectare, which even if an error range of +/-1 magnitude be allowed isn't very much: Consider that even if each nodule fixes 10 times its own weight a year (this figure is pure speculation) results in 0.5-0.6 kg per hectare which is two magnitudes less than N fertilizer ratios used in most nitrogen fertilization experiments - to put it in more optimistic words, there is still much room for improvement.
6.3.1. N fixation: The pessimistic opinion of the preceding section might be verified or invalidated by sending dried legume DM (each of the more common ones 5-10g) and a lot of non-legume DM(ca. 1kg) collected in the transect experiments, it might be interesting to send them to a lab specialized on natural 15N-isotope abundance fractionation (proposed by #Doughton 1992, applied by #Michelsen 1996).
6.3.2. Indigenous legumes: As pointed out in some detail in section (2.3.), some of the indigenous legumes might be used for dune revegetation, medicinal uses or be controlled by insects. The exact distribution data provided by this study should be helpful for this.
6.3.3. Rhizobia: As pointed out before, Onobrychis nodulation was relatively poor and there might be a potential to improve its capacity. However, Onobrychis does not come to viable seed formation and annual medium temperatures of 1.1oC are probably a bit too low for Onobrychis cultivation (#Duke (1981) recommends annual medium temperatures above 5oC). Furthermore, the large-scale production of fodder is -at the moment- not very feasible for a variety of socioeconomic reasons, and a more basic step (to be done first) is to convince the local Tibetan nomads of its usefulness, which will only work if the approach will "pay" under market-economy conditions. Another productive step to overcome some cultural barriers would be some ethnobotany on the use of and the popular opinion on some plant species in the area. In the meantime, it might however be interesting to include our strains in nationwide or international collections, which would make them available for theoretical and applied research in other provinces or countries. For example, they might prove valuable for the already more advanced agricultural systems in Gansu or Inner Mongolia.
An area more promising than grassland rhizobia is to work on Sichuan plain rhizobia, as here N fertilizers are indeed used in huge scales. It might be fruitful to this in close cooperation with other rhizobial research units such as the Sichuan Agricultural University (Yaan) which is the most active center for rhizobial research in our province.
6.3.4. Numerical techniques: It
should be interesting to reevaluate the results of previous
phenotaxonomic studies by the algorithms and programs presented in
section 4 as well as to screen their results for selective media as
proposed in section 5.