Created: 1997-01-07, Last update: 1997-03-30, Author: Holger Blasum, URL: http://www.blasum.net/holger/wri/alpleg/engl/ch1.html, Parent: http://www.blasum.net/holger/wri/alpleg/engl/index.html

1. Useful background knowledge for legume N fixation research in a cold grassland area like Xiaman

Abstract: This introduction covers physiology, the Xiaman experimental site, N fixation ecology and research methods. Its aim is a) to explain basic ideas to a reader with a background in general biology and b) to give access to recent literature in the fields concerned.

1.1. Physiology of general N fixation and the legume-rhizobium symbiosis

1.1.1. Evolution and energetics

Nitrogen, element No. 7, is an essential element for all forms of life so far discovered on our planet, it is for example part of all proteins and nucleic acids (the carriers of genetic information). Unfortunately, in the oxidative atmosphere prevalent since ca. 2 billion years, most nitrogen has not been readily available to most organisms, but rather been concentrated as dinitrogen gas in the atmosphere. Although the covalent N-N bond is rather stable, a number of organisms, scattered throughout the kingdoms of the Eu- and Archaebacteria (#Woese 1987), have parallely acquired the ability to fix nitrogen, maybe relatively lately in evolution by horizontal transfer of genetic information(#Postgate 1992). Nitrogen fixation, i.e. the transformation of atmospheric nitrogen into hydrogen- or ultimately carbon-bound nitrogen which all organisms can use, proceeds via enzymes called nitrogenases (quite a misnomer), the entire structure of the most efficient molybdenum variety believed to have been uncovered recently. It consists of different Fe/S clusters, an ATP-reaction site and an active Mo site for dinitrogen binding and reduction. As the molybdenum center is activated by hydride formation, one necessary side-product of the reaction is hydrogen (possibly protecting the oxidation-sensitive enzyme). In legumes, special hydrogen recycling (uptake) systems exist. It should be noted that nitrogenases with non-molybdenum coenzymes such as vanadium-iron in the azotobacteria are considered to be less efficient because they release huge amounts of hydrogen (#Frausto 1993:426).

Of course, this ingenious way of capturing atmospheric nitrogen is not free, but rather energy-intensive. #Rao(1987:247) reviews estimates between 1.26-1.6 mol glucose consumed per mol of nitrogen fixed into carbon skeletons. However, this doesn't look that horrible when compared to the costs of nitrate incorporation (1.4-1.5 mol glucose consumed per mol of nitrogen fixed) or even ammonia assimilation (also consuming 0.7-0.8 mol glucose, the bulk of it used for carbon skeleton incorporation). Or if I look at it from a monetary viewpoint, an estimated 150-175 billion kg annual biological nitrogen fixation worldwide (Rao, ibd.) equal ca. 1.2-1.4 trillion (US $ 160-180 billion) Yuan of equally if calculated after Chengdu market prices for urea fertilizer.

Being so useful, nitrogen-fixing microorganisms often happen to end up in more or less close symbioses, such as the Frankia actinomycete with many (especially woody) plants (#Benson 1993#Forestry Soil Institute of the Academia Sinica 1985#Deng 1992) or cyanobacteria in lichens or with the tiny Azolla on rice paddies or azotobacter in the cow rumen. The best-investigated of these communities is the legume-Rhizobium symbiosis, which shall be briefly introduced below.

1.1.2. Particulars of the legume-rhizobia system

1.1.2.1. Symbiotic partners

Legumes ("bean plants"/Fabaceae) are a family or superfamily of dicotyledonous plants, in species diversity they are third most large family of plants (ca. 5% of Earth's plant biodiversity, after the Poaceae("grasses") (7%) and Orchidaceae(7%)). It may be safely guessed that they make up for more than at least for 1% of the total Earth plant biomass, and they comprise well-known crops such as soybeans, mung beans, faba beans, French beans, peanuts and peas as well as forages such as clover, sainfoin, fenugreek and vetches which ranks them second in importance of nutrition for mankind (after the Poaceae; see #Allen 1981for extensive account; for a database (#Bisby 1996) visit www.soton.ac.uk/~ildis). Although during the "Green Revolution" advances in legumes breeding haven't been so big as in the Poaceae (definitely mankind's most important nutrient source for the last five millennia), they are now being reappraised as a useful tool for ecological agriculture (#Giller 1996).

The soil is probably the most abundant habitat for microorganisms, and this is also the place where rhizobia do thrive (although they tend to be present in low numbers). For bacteria doing such energy-intensive jobs as nitrogen fixation, the area around plant roots ("rhizosphere", #Yan 1993 #Doebereiner 1988) is particularly attractive. So it doesn't come as a surprise that most rhizobia preferentially interact with the legume("bean-like plant",see below) root, and this is the habitat their name is derived from. From their phylogenetic relationships (being in the alpha-proteobacteriacea/alpha-purple bacteria, #Stackebrandt 1988) it is evident that, similar to mitochondria, phyllobacteria and agrobacteria, they have been a plant pathogen, but one which has been tolerated due to its usefulness for the host plant.

The symbiosis is confined to a special organ called nodule (meaning: "little knot") in most Western languages, referring to the typical knot structures visible on most temperate grain legumes in Europe (such as peas). Interestingly, the retranslated Chinese idiom ("genliu") means "root ulcer" which corresponds to the more ulcer-like form on Asia's dominant grain legume, the soybean. When dwelling on East-Western nodulology, it might be added that there is a vague speculation that the ancient Chinese character for "bean" included three dots indicating nodules (#Hymowitz 1970 #Bergersen 1980), but this theory which would antedate the first Western documentation (Fuchs 1542) by more than two millennia doesn't seem to find much credit in China (try to look up the Hymowitz paper in a mainland library).

For a long time, rhizobia have been believed to be able to fix nitrogen only in the nodules, and it was as late as 1975, that #Pagan demonstrated rhizobial N fixation on an N-free and legume-free medium. Although more recent findings (such as #Dreyfus 1983) affirmed that the capability of independent N fixation is widespread, it is still believed that the bulk of rhizobial N fixation occurs in the more eloborate nodule system.

1.1.2.2. Different legumes often need different rhizobia

19th-century microbe hunters can be credited to have classified nodule organisms even before anybody opined they might be useful (#Schroeter 1886, see #Moffett 1968). Within a decade of the discovery of the nitrogen-fixing properties of the symbiosis (the classic paper is #Hellriegel and Willfarth 1888), it was established that not every kind of bacterium will be able to "nodulate" (i.e. to form little knot-like structures on) every plant. This led to the formulation of a dozen of "cross-inoculation groups" (#Mueller 1925#Fred 1932), these are groups of strains which are capable of replacing each intragroup strain in order to form symbioses with a certain group of legumes, but strains from different groups are unable to replace each other. For examples, vetches and peas form such a group, and if you isolate a strain from faba bean, you might expect it to nodulate the green manure vetch Vicia sativa, but not soybean which is in a different group. To give another analogy, Chinese people are able to read Chinese books (regardless of published in Harbin or Haikou), Japanese are able to read all material published between Hokkaido and Okinawa, and so on. But you will note that this correlation is not a perfect one: quite a few Chinese know Japanese and vice-versa. Furthermore, in China, there are millions and in Japan there are thousands whose mother tongue isn't Chinese or Japanese at all. So this relation (although of great practical value) is a limited one, and the same is true for rhizobial inoculation groups, some groups are rather well-defined (such as the above-mentioned vetches or medics) and others are indeed quite loose; however there is no need to be as cynical as #Wilson (1944) who polemically termed his paper on astragalus rhizobia "Over 500 reasons for abandoning the cross-inoculation groups of legumes".

1.1.2.3. How different legumes recognize different rhizobia

Although - in great contrast to section 1.1.2.2. - the details of the plant-rhizobium interaction are not at all crucial for the understanding of this thesis, this fascinating model for symbiotic interspecies communication deserves to be introduced - the major steps of the most common pathway are:(1)If there is a need for nitrogen (#Coronado 1996), the legume produces flavonoids and excretes them to the rhizosphere.(2)Water-soluble flavonoids travel through the soil, rhizobia meeting them can dissolve them by C-ring fission(#Rao and Cooper 1994).By a yet unknown mechanism, a nodD gene is activated by flavonoids to produce a NodD protein activating other nod ("nodulation") genes (#Spaink 1994),(3)the most important being nodABC(all species of rhizobia so far investigated have them,#Tan and Chen 1992). NodABC can produce the skeleton of the Nod factor eliciting nodules. The nucleotide sequence of NodC resembles a chitin polymerase, and possibly catalyzes N-acetyl-D-glucosamine polymerization. After tetramers or pentamers of chitin are built, NodB removes the apical N-acetyl-glucosamine residue, after that NodA catalyzes the addition of a longer ester(#Denarie 1993).Although rhizobia can excrete Nod factor into the medium, from its chemical structure it a position on the cell membrane seems more likely(#Hirsch 1992).The chemical structure of the Nod factor of each kind of rhizobia is different, broad host-range rhizobia (such as NGR 234) have several Nod factors (#Franssen 1995). The first nod factor whose structure has been cleared up was that of alfalfa(#Lerouge 1990).Except for the common nodABCD genes, many nodulation genes are species specific, e.g. alfalfa's NodH, NodP and NodQ are responsible for the sulfurylation of the Nod factor;NodE,NodF and NodL participate in the fatty acid synthesis(#Caetano-Anolles and Gresshoff 1991). Alfalfa rhizobia lacking NodH have a Nod factor which is not sulfurylated,thus they cannot nodulate medics, but are able to nodulate vetches and peas(#Horvath 1986 #Denarie and Cullimore 1993).

When the rhizobial Nod factor comes to the plant root in nano- or picomolar concentrations,the classical reaction is:(4)first a calcium protein (rhicadhesin,#Smit 1987) shared by all rhizobia and agrobacteria mediates binding to the plant receptor having an Arg-Gly-Asp domain,possibly a lectin(#Swart1994) . For example, Arg-Gly-Asp is also distributed in a lectin called discoidin (#Barondes 1988). Of course, rhizobial binding is also influenced by environmental factors, such as pH,Ca,Mg and phosphate availability (#Caetano-Anolles 1989#Howieson 1993). (5) After 6-18 hrs, the fine root hairs of rhizobia begin to curl(#Hirsch 1992),(6)the infection threads are formed in the root hair(7)Rhizobia penetrate via the infection thread to the root cortex, and induce the cortex (inner cortex for indeterminate nodules,outer cortex for determinate nodules) to produce a meristem. Some researches think this step requires rhizobial exopolysaccharides, but this is still unclear.(8)Controlled by auxins, this meristem produces nodules( #Hirsch 1989),a new organ. In the well-researched alfalfa, this can be divided into an apical meristem, a medium zone containing rhizobia-containing plant cells (very inappropriately named "bacteroids"), and basal vasculae (tubes) maintaining contact with the plant metabolism.(9)For aeriation, the plant produces leghemoglobin and several other enzymes(#Caetano-Anolles and Gresshoff 1991).(10)The host plant can control the total amount of nodulation:#Nutman(1952)discovered that if you cut away the nodules of red clover,it is able to regenerate nodules, but will not form new nodules without cutting. This phenomenon has been discovered on quite a few plants, but its molecular biology is still unclear.(11)The nodules of some plants are shed in winter, some appear to be perennial. There has been little research on the mechanisms of this.

1.2. The Xiaman experimental site

1.2.1. Location on the Qinghai-Tibet plateau

The Qinghai-Tibet plateau (comprising nearly all of Tibet, whole Qinghai, parts of Xinjiang, Gansu and Sichuan and the alpine regions of most adjacent Himalayan states) is the hugest high plateau of the world, and being the cause of the monsoon winds it has profound influences on the climate of adjacent South East Asia. But although a unit geologically and culturally, due to different moisture belts, its vegetation should be divided into subunits ranging from coniferous forests in the southeast to desert belts in the northwest. Our research area is located on the Naqu-Yushu alpine-semihumid belt (which is a bit colder and moister than the better-known Lhasa shrub grasslands), which has been characterized by #Zheng (1979) as following:

"This area is located in the center and east of the plateau, it extends from the east of the Nujiang (Mekong) via Yushu, Guoluo to Ruoergai in NW Sichuan. The relief is rather shallow, there are broad valleys, basins and hills, the altitude is about 4,000-4,600m, in the eastern Ruoergai down to 3,500m a.s.l. The glacial relief is developed, and there are islets of frozen soil surviving. The mean temperature in the warmest month is 6-10(12)oC, the annual precipitation 400-700mm, dryness 0.8-1.5, in summer hailing is common, and snowfall in winter is not negligible. In the high alpine meadows and shrublands Kobresia, Polygonum, Salix and Rhododendron are dominant, and create a high alpine felt soil. Pedogenesis is characterized by large accumulation of humus, little washing out, and long-time glaciation. As the grass layer is quite hard, solifluction processes such as landslides and debris flows are common. River systems are developed and peat soils do often form."

Most of the plateau is used for cattle grazing by Tibetan nomads (for an English introduction see #Miller 1990), and it is claimed that since the reprivatisation of grazing systems overgrazing has led to deterioration in some areas (#Xu 1990): For example, in Qinghai province there area per sheep unit has been estimated to drop from 1.8 ha to 0.7 ha in 1985; meat weight dropped by 30% per individual and the whole productivity of wild grasslands dropped by 30-60% and with an desertification rate of 1.8% p.a.(#Shi 1992). The detrimental effects of overgrazing have also been shown by #Han et al.(1991) who reported an increase in forbs from 21.7% to 61.6% and an increase in poisonous plant biomass from 0.57% to 1.39% when cattle density was increased from 2.14 to 6.07/ha.

Furthermore, although to date relatively unspoiled by pollution, the plateau might suffer from global warming; a joint State Meteorology Administration (SMA) - WWF study predicts this environment to shrink about 28% to the middle of the next century (this is a 1992 estimate, see #Feng and Wang 1996).

1.2.2. The Xiaman environment

In the eastern part of the Qinghai-Tibet plateau, the characteristic peatlands in the provinces of Tibet (#Zhao et al. 1982), Gansu (#Yun et al. 1994), and especially NW Sichuan (#Qi 1960,#Chai et al. 1963,1965,#Zu 1983,#Yang 1986,#Sun et al. 1987,#Tsuyuzaki 1990,#Tsuyuzaki 1992, #Bjoerk 1993 #Yang and Jin 1993B, #Zhao1995 and 1996) have attracted much scientific attention, but data on other vegetation types (such as the subalpine hill vegetation) is more scarce (#Liu 1984#Ni and Wei 1984#Yang 1987#Chen 1992#Zhang 1994 #Liu 1994). The most exhaustive account on Xiaman vegetation has been given by #Wei Taichang and Zhao Zuocheng (1986), a species list of vascular plants and vertebrates can be found in the appendix (on disk) of this work.

1.2.2.1. Relief and climate

The Ruoergai high plateau (sometimes included in the "Songpan grassland") is a plateau surrounded by mountains of ca. 4000m a.s.l., extends over 200km NS direction and 100km EW direction, its average elevation is more than 3,400m. The rock formations are rather simple, mainly consisting of Triassic graygreen or yellow sandstone (most of it metamorphosed to slates and phyllites) and black shales. The relative height of the surrounding mountains is 300-500m, but the hill height inside the area is usually 70-150m, sometimes up to 200-300m and running parallel. Most SE slopes are rather steep and cirques rather complete, NW slopes are more genteel. River valleys are very broad (up to 18km) and have many sediments, at areas where waterflow is sluggish marshlands form (#Chai et al. 1965).

Although in winter there is not much precipitation, the area is moister than most of the middle and western parts of the plateau: in the semiarid winter climate changes to semihumid in May and thanks to Indian and South China sea monsoon it humid from June to October; so according to the Penman formula its aridity is less than 1 (0.93) and belongs to the humid zone (#Qian and Lin 1965), so the turning green and wilting times are earlier than in the rest of the grassland (#Zhang Yiguang 1985). Inside the Ruoergai grassland, the north is more slightly more arid than the south, data for Longriba are 753mm (#Chai and Jin 1963), Hongyuan 613mm (#Qin et al. 1985), Tangke 647mm (#Chai and Jin 1963), Ruoergai 657mm (#Ministry of Civil Affairs 1993), Maqu 616mm (#Yun 1994).

Mean annual temperature data are much more uniform, viz. 1.0oC for Longriba (#Chai and Jin 1963), 1.1oC for Hongyuan (#Qin et al. 1985), 0.8oC for Tangke (#Chai and Jin 1963), 0.9oC for Ruoergai (#Ministry of Civil Affairs 1993), 1.1oC for Maqu (#Yun 1994). Most published data available is on Hongyuan, its medium temperature in July is 10.9oC, diurnal temperature amplitude is 16.1oC, annual solar irradiation is 2147 hrs., in the growing season photsynthetically active irradiation is 31000 J/cm2*month during May-Aug and 22000 J/cm2*month during Sept-Oct, photosynthetically fixed energy is less than 1% and biomass produced yearly is 2000-3000kg/ha in most grasslands. Average peat soil temperatures at 5 cm depth are 5.9oC in May, 13.6oC in July, 9.3oC in September. From end of October to end of April the soil freezes, reaching a maximum ice depth of 40cm (marshlands) and 60cm (grasslands) in February (#Yang and Jin 1993).

1.2.2.2. Soil and vegetation

Soils are mostly influenced by water availability and according to the "diagnostic" Chinese soil classification system (#Gao 1990;note that there is some discussion to modernize the soil classification system,see #Wang (1994)) can be classified into three forms (#Chai et al. 1965):

(I)Organic soils: peat soils: mainly distributed in the broad river valley of Heihe and partly distributed in Baihe river valley, continuously, seasonally or temporary waterlogging occurs. The peat layer usually is more than 3m thick and can occasionally reach 6m. These soils are very rich in organic matter (more than 50%), total nitrogen (5-8%), the pH is 7.0-7.8. The communities (I.I) and (I.II) have similar nutrient dynamics, crude protein is highest in June, biomass highest in July/August (1500 and 1650kg resp.).

(I.I)Marshlands always moist: Equisetum limosum, Potamogetum, Hippuris vulgaris, Myriophyllum spicatum,Cicuta virosa, Menyanthes trifoliata, Glyceria aquatica, Polygonum aquaticum,Triglochin maritimum, Carex meyeriana, Utricularia media etc.

(I.II)Marshlands seasonally moist or intermediate stages:Carex muliensis, Carex capillifolia, Sanguisorba filiformis, Caltha scaposa, Aster alpinus, Deschampsia caespitosa, Blysmus sinocompressus,Gentiana sino- ornata, Chamaesium paradoxum, Ranunculus longicaulis/pulchellus,Trollius ranunculoides, Kobresia humilis/parva/tibetica, Parnassia trinervis, Koeleria cristata, Juncus leucanthus,J.concinnus,Pedicularis resupinatus,P. rhinanthoides, Ranunculus brotherusii var.tanguticus, Anemone obtusiloba, Eleocharis valeculosa, Koeleria cristata,Poa chalarantha,Elymus nutans (#Bjoerk 1993). In case of overgrazing, the tall grasses (such as Poa, Elymus) will become less and the cyperaceae will dominate. In areas disturbed by rats, tuber root plants such as Blysmus sinocompressus,Carex enervis,Potentilla anserina will have an advantage.

(II) Alpine brown soils (or: alpine yellow-brown lime soils, ("Sichuan Land Resources Map Collection" 1990), sandy grassland mixed soils ("Hongyuan Soils" 1985). On the dry river banks of Bai river (such as Tangke) and the Reerba in Ruoergai. The matrix is yellow sediment soil, the upper 0-4cm are close root cover, then follows are brown-black humus layer (30-40cm thick), with an organic matter content of 3-5%, it has carbonic acid inlayers (up to 10%), pH 6.0-7.5, below 40cm it becomes sandy, sometimes there are small dots of ferrous inclusion. This soil is the most fertile in the area.

(II.I) The Xiaman river plain sandy soil environment has been characterized by Liu (1994) who listed Elymus nutans, Elymus sibiricus, Geranium pylzowianum, Kobresia capillifolia (=maquensis), Anemone rivularis and other plants.

(II.II)In overgrazed grasslands(plant height 10-30cm)the dominant plants are: Stellera chamaejasme, Artemisia hedinii, Aconitum carmichaeli, Pedicularis remitorta, P.oederi, Elsholtzia densa, Eruca sativa (#Yang Dingguo 1987). In dry areas, Stipa spp. will dominate.

(III)Subalpine felt soils(also called: "subalpine grassland soils"; "histosols" on the 1974 FAO World Soil Map):Distributed on the surrounding subalpine mountains and hillocks (3420-4000M; alpine grassland soils in the highest points). In contrast to category (II), the matrix is sandstone shale or slate. The grassroots layer is several cm, the brown-black humus layer 20-30cm deep and organic matter content reaches 5-15%, pH is neutral.

(III.I)Here we find typical subalpine "Five-Flower-Meadow"/ forb meadows with a height of 30-80CM,representative are :Kobresia setchwanensis, K. capilliforma, Elymus nutans,Roegneria nutans, Koleria litwinowii, Helitrotrichon tibeticum,Brachypodium sylvaticum,Agrostis spp., Carex moorcroftii,Carex filipes, Stipa aliena, Polygonum viviparum, P. amatum, P. avicluare, P.sibiricum, Saussurea graminea, Anaphalis hancockii, Anaphalis lactea,Swertia franchetiana, Gentiana timensis, G. straminea, Ranunculuspedicularis, Festuca ovina, Clinelymus sibiricus, Trollius ranunculoides, Leontopodium longifolium, Allium cyaneum, Scrofella chinensis, Coluria longifolia (#Liu Qi 1984 specimen by Zhang Zhaoqing).

(III.II)On southern slopes, xerophilous plants such as Anaphalis,Stipa spp., vertches. Kobresia spp. will become more dominant.

(III.III)On northern slopes, typical shrubs include Spiraea schneideriana var. amphidoxa, Potentilla fruticosa, Spiraea alpina, Lonicera tibetica. Above 3800m Rhododendron violaceum can be found. Herbaceous plants to be found includeHelictotrichon tibeticum, Festuca rubra, F. ovina, Elymus nutans,Brachyelytrum erectum, Poa annua, Koeleria cristata, Ptilagrostis dichotoma, Deyeuxia scobrescens, Carex digyna,Polygonum viviparum, P.sphaerostachyum, Pyrethrum tatsiense, Hedysarum sikkimense,Meconopsisspp.(#Liu Qi 1984). On degraded slopes, these are replaced by shrubs like Caragana tibetica,C.erinacea(height:45-50CM)and herbs like Elsholtzia fruticosa, Festuca aliena, Kobresia pygmaea,Saussurea sp.,Potentilla multicaulis, Poa annua and Iris spp.(#Chen Quangong 1992).

(III.IV)On moist riverbanks Rhododendron violaceum, Artemisia spp. , Salix spp.,Hippophae can be found.

(III.V)On the upper levels of the surrounding mountains, from 3800-4000M to 5200M alpine felt soilswill become prevalent and their typical vegetation includes Kobresia pygmaica, Poa sinattenuata var.vivipara, Polygonum viviparum, Coluria longifolia,Spenceria ramalana, Anemone geum, Anaphalis flavescens, Leontopodium longifolium, Saussurea stella, Arenaria sp., Androsacetapete etc.(#Liu Qi 1984).

(IV) Parallel to the yellow river, dunes can be found. They have increased in height from 2-3m since the 60s (#Chai 1965) to 5-10m, their plant cover is scarce, however rich in legumes. See section 2.3.5. for discussion.

1.3. Ecology of leguminous nitrogen fixation

1.3.1. Population dynamics

Rhizobia are relatively scarce, their number seldom exceeds 1% of the soil population. If let alone in the soil, their numbers decrease (#Cao 1994): they are rhizosphere organisms highly stimulated by the presence of plant roots, in the short term these stimulating plant roots are not necessarily legumes, on the long term however presence of a certain host certainly enhances the presence of appropriate inoculation group rhizobia: #Mahler (1982) reported a rhizobia-other bacteria ratio of 1/50000 is non-cultivated and 1/500 in cultivated alfalfa fields. In a tropical soybean-rice rotation environment, #Simanungkalit (1995) found population density changes of 1-3 magnitudes between plant cultivation and fallow periods, and these changes apparently do not rely very much on which plant is cultivated: for example, he counts 72 rhizobia/g soil in a fallow lowland rice field and 26000 rhizobia/g for lowland fields cultivated with rice. In a temperate environment, #Wiehe (1995) showed that marked rhizobia grew nearly as well as pseudomonads in maize, wheat and rape rhizospheres. He also reported that during one growth period rhizobia were able to colonize non-legume rhizospheres as far as 0.6m apart from an original site of inoculation.

1.3.2. Habitat-dependent efficiency: crops and pastures

In crop environments legumes are either used as cash crops or as green manure fertilizers. Although green manures such as Astragalus sinicus and Sesbania rostrata have an efficiency equal to an average fertilizer dressing (both being able to fix 80-100kg N/ha in 45-60 days; #Becker 1995) due to present cheap urea fertilizers the manure use is declining at the moment in a world-wide scale. In the Sichuan basin green manures that can be occasionally found in marginal areas are Astragalus sinicus and Vicia sativa.

Still, legumes make up for a reasonable share of crops and they can be nearly as effective as explicit manures. In subtropical humid fertile environments such as the Sichuan basin around Chengdu, legumes that are used directly as food are faba bean and French bean (growing all over they year, main harvest in May), mung bean, cowpea, soybean, Dolichos lablab and peanuts (harvested in September). All of these species are relatively sturdy and due to huge plant size also have huge nodules. For example, with Chengdu Dolichos lablab nodules of a diameter up to 3cm can be found. The products of nitrogen fixation go directly into the plants and their seeds. Of course, this sometimes is responsible for the fact that contrary to popular belief some legumes extract rather than increase soil nitrogen content (#Rao 1981249), however adverse effects can be ameliorated if legume stover is returned to the fields: #Toomsan (1995) reports N fixation rates of 150-220kg N/ha for peanuts and 100-152kg N/ha for soybeans in Thailand, and if the stover is left on the field, after harvesting the N-rich crop some 13-100 kg N/ha are left in a peanut field; however soybean cultivation resulted in a net loss of 37-46kg N/ha. In comparison, a crop like maize usually takes out some 150kg N/ha.

Even in very cold environments as Murmansk (north of the Polar Circle), where nodulation on wild Astragali and Oxytopis were considered as unsatisfactory (#Roizin 1959), a pea crop were able to fix 28-76 kg N/ha (#Egorov 1985 measured by ARA) and on the Qinghai-Tibet plateau (Lhasa area between 3500-4100m altitude, ann. mean temp. 6-8oC), crop rotation systems with barley, wheat, peas and faba beans (#Zhu 1990 #Zhou 1991) have been achieved.

In temperate grasslands (such as at Xiaman) legumes occur as wild plants, they seldom attain the large size of crop legumes (#Zhang Xiaochuan 1989). They also have a high protein content, but as they usually don't make up a very large proportion of plant biomass, secondary effects such as nitrogen conversion might be more important, this can happen by decomposition of legumes and nodules and has been estimated at 3 to 102kg N/year*ha or 2-26% of biological nitrogen fixation for clover and alfalfa pastures (#Ledgard 1992#Thomas 1993).

On the Qinghai-Tibet plateau, #Yang (1995) in Hongyuan also reported a poor nodulation for high-altitude wild legumes and #Nie (1989) in Gansu achieved good results with medic inoculation. Obviously the amount of leguminous nitrogen fixation is also dependent on many factors, such as temperature, soil pH and soil nutrition, salinity, soil moisture etc. In the following some factors possibly important for Xiaman will be discussed.

1.3.3. Temperature All leguminous host plants so far investigated have a normal Calvin photosynthetic cycle, with optimal temperatures of 15-25oC (#Bordeleau 1994). Most rhizobia are thought to grow best at a temperature of 28 to 31oC (and 28oC is indeed the standard temperature for culturing rhizobia), but S. meliloti (medic rhizobium) grows well at 35oC, and arctic rhizobia grow at 5-10oC (#Graham 1992). Obviously, the optimal temperature of the symbiosis is highly dependent on the environment it is found in: while for subtropical regions #Lie(1971) reported that a pea cultivar Iran was nodulated at 26oC by R.leguminosarum, but not at 20oC, #Bordeleau (1994) determined the optimal temperature for an Oxytropis maydellana - rhizobium synthesis at 15-25oC. The Canadian research group also found that at 10oC arctic rhizobia grew relatively faster than temperate ones, and nitrogenase activity was detectable at up to -4oC (ibd.), whereas in alfalfa below 8oC nitrogenase and nodulation cease (#Cralle and Heichel 1982). One should keep in mind that nonsymbiotic plant nitrate assimilation is inhibited by low temperatures as well (#Atkin 1994).

The evidence that temperature optima for the symbiosis are closely related to its natural origin must be taken in consideration when culturing the symbiosis in nodulation studies.

1.3.4. Acidity

Rhizobia, especially fast-growing rhizobia (#Cao Jingle 1994A & 1994B, except R.loti and R.tropici, #Graham 1992), are more sensitive to acid than most other bacteria, which can be easily demonstrated on agar plates. In the field soil, this sometimes turns out to be a matter of practical concern, especially for S. meliloti: #Brockwell (1991) found that in Australian soils at pH=7.0 there are about 89,000 wild medic rhizobia/gram soil, but only 37/gram soil at pH=6.0. More disconcertingly, #Evans (1980) working on peas found that although legume and rhizobia do tolerate a certain level of acidity, nodulation was 10 times more susceptible to acid than soil survival of each of the partners and inoculation experiments with acid-resistant strains have been successful in the St. Petersburg area (#Fesenko 1995). At very low pH (below 5.0) excessive quantities of aluminum and manganese will be liberated below. This is not a major problem on the grassland, but might be of relevance for some areas in the Sichuan basin, where soil pH is below 5 (such as yellow soils around Chongqing with a pH 4.4-4.6, see #Feng 1996 or the Longmen Mt. SE of Chengdu). Although relatively harmless to the rhizobia, aluminum and manganese stunt plant root growth and thus influence nodulation, a problem which can be overcome by either plant cultivar selection or soil liming (Bordeleau 1994).

On the other hand, though most rhizobia which maintain a slightly alkaline intracellular pH (#Glenn 1994) are thought to be most efficient at neutral pH, #Tang and Robson (1993) found a pH above 6.0 to reduce nodulation in lupines.

1.3.5. Soil nutrients

The soil nutrient distribution is most intricately related to pH levels: calcium, phosphate and molybdenum become scarce at acid pH, and reports can be found about the stimulating influence of each of them (for Mo, see #Tu 1992, for Ca and P, see #Graham 1992). Legumes are more rich in potassium and phosphate than most other plants, and interestingly phosphorus seems to accumulate in the nodules as well as rhizobia (ibd.). So quite a lot of work has been done to combine nitrogen fixation with P-accumulating mycorrhiza (symbiotic fungi, #Buttery 1992#Lynd 1995).

A high content of rapidly available soil nitrogen induces some plants to suppress nodulation (by stopping flavonoid excretion, see #Coronado 1996). This plant autocontrol of nodulation (#Caetano-Anolles 1991) thus averts wasting energy when enough nitrogen can be scavenged, which can be undesirable for the breeder or pastoralist, but if this problem occurs it can be avoided by choosing appropriate cultivars. For the majority of wild legumes it is still unresearched in how far they are influenced by this.

1.4. Research methods

The structure of this section basically follows the classic books on methods by #Vincent (1970) and #Bergersen (1980); Chinese translations of both works have been published and are widely available. So the emphasis in this presentation is on what is new (not mentioned in those works) or especially relevant for this thesis.

1.4.1. Determining overall legume quantity and distribution

In natural ecosystems the first step is to ascertain whether there are wild legumes and to get some basic data about their biology. Host plant identification should be correct or at least reproducible (deposit specimen in the local herbarium). Standard methods for assessing biomass dynamics and plant communities can be found in #Jiang 1988

An easy and objective way of sampling is along a transect, in plant biology the most "classical" example is probably the Californian transect experiment by #Clausen,Keck and Hiesey(19401945,1948,1952;reviewed by #Davis and Heywood 1963) crossing two mountain ridges (with a relative altitude difference of more than 2000m) for several hundred km, they investigated the distribution and environmental requirements of multispecies forb genera like Sisyrinchium (Iridaceae), Aster, Artemisia, Achillea,Horketia (Asteraceae) and Potentilla (Rosaceae). Transects of any length have been used to assess very diverse biological phenomena, such as insect (#Zhou 1992) or fish distributions (#Hong 1989). Very recently, #Fang and Ohsawa (1996) published data of forest distributions along the 30N transect in East Asia (transect length: more than 1/8th of the 30N meridian!).

Obtaining these data is laborious, but often quite low-tech, and the more difficult problem is to analyze them appropriately and to filter out relevant relations. As totally unknown plant distribution data should not be assumed to be normally distributed, for this it is recommended to employ more robust non-parametric statistics such as Kendall's tau, the Wilcoxon-Mann-Whitney rank tests or Chi-square tests (#Wall 1986#Sprent 1989). From the statistical point of view, to discover a non-obvious, but not very hidden relation (by nonparametric statistics) a number of 10-100 samples is usually optimal.

1.4.2. Assessing N fixation

The crudest approach for evaluating nodules in the field is to observe their color: as effective nodules contain leghemoglobin to ensure the aereation of the nodule-enclosed rhizobia, nodule color gives a first hint on efficiency, white being "bad" and red being "good". Of course, these qualitive observations are not very amenable to quantification, and so a variety of more exact methods has evolved.

1.4.2.1. Direct approach: acetylene reduction assay

Nitrogenase is an enzyme which is not strictly specific: besides the reduction of nitrogen it also readily catalyzes the reduction of acetylene to ethylene. As both compounds are gaseous and both are not part of the natural atmosphere in significant quantities, this is amenable to gas chromatography analysis, and from the amount of ethylene measured inferences about the nitrogen fixation can be made. This has been a standard method BNF assessment during the 70s and 80s, and for example in our province #Deng (1992) used it for assessing non-leguminous tree N-fixation. However it has the serious drawback that it requires swift movement of samples to the lab (not feasible in Ruoergai) or a field-mobile gas chromatograph. Furthermore, the ARA N fixation data are just a snapshot (usual measurement times are several hours) and thus neglect that overall N fixation is in a complex homeostasis between many biotic and abiotic factors (local N content, development stages of individual plants, temperature changes etc.) and different ARA data are not always easy to compare.

1.4.2.2. Indirect approach: N labeling and depletion studies

The N content of any matter can be determined by the Kjeldahl analysis (sulfuric acid digestion), however although legumes usually have a higher N-content than non-legumes, absolute N contents do not tell us much about its origin. However, different isotopes of N allow to enrich an fertilizer with radioactively labeled N, and to calculate the ratio of radioactive 15-N (non-fixed) to non-radioactive 14-N (fixed) nitrogen. Soon this technique evolved to a N depletion assay (#LaRue and Pattison 1981#Xie 1991) taking advantage of the fact that 0.3663% of atmospheric nitrogen is actually radioactive, so that instead of hyperradioactive fertilizer hyporadiactive could be used; with this assay the N-fixing plants become more radioactive than their fertilizer-fed counterparts. With increased precision of analysis this again has evolved into an 15-N isotope fractionation technique (#Ledgard 1992#Doughton 1992 for application see #Bolger 1995and #Michelsen 1996); this method does not use any fertilization, but rather directly calculates N fixation in the field from the ratio of radioactive to non-radioactive nitrogen.

1.4.3. Assessing N demand

This is conceptually very straight-forward: make a fertilization experiment and see whether there is any response in biomass production. However, N application should not be too early as otherwise nitrogen will have been lost as ammonia. Attention should be paid whether the fertilization has any adverse effects on nodulation patterns.

1.4.4. Inoculation techniques

For long times it has been known that, when a new legume is grown on an unknown soil, sometimes good results are achieved by inoculating the soil with soil from different area where the crop has been grown for a longer period of time. With the discovery of the N fixing organism it has then become possible to apply rhizobia directly to soils, the earliest well-known product, Nitragin, going into production in 1896, that time consisting for pure bottled rhizobia broth (#Smith 1992), and subsequently a series of more sophisticated techniques evolved. However, as it is very easy to get a microorganism into the soil and very difficult to get it out of it, non-native strains must be thoroughly tested for their effectiveness.

1.4.4.1. Isolation and maintenance of rhizobia

As rhizobia are concentrated in root nodules, finding rhizobia is relatively easy: just dig out a plant and smash it's cleaned nodules in a sterile environment onto on agar plate. However, this of course only finds those rhizobia that most competitively nodulate plants in the field and no latent (possible more effective, but less competitive) strains. Thus recently a variety of DNA-DNA hybridization (#Louvrier 1995) and PCR-based (#Pillai 1992) identification methods have evolved to isolate rhizobia directly form the soil. These researches have led to a recent reawakened interest in selective media, some of which are used for isolating, others for checking of rhizobia.

Unfortunately, earlier reviews (such as #Mueller 1925and #Fred 1932) weren't very optimistic on this selective media. Moreover, some media reported in the 60s and 70s (such as #Graham 1969) were subsequently denounced as irreproducible (#Pattison 1974) or too inhibitive (#Bromfield 1993). Thus it was only very recently with the advent of reliable molecular strain identifaction methods (see section 1.4.4.2.2.) that serious interest in agents selective for rhizobia has reawakened (since the early 90s). Most of these reports center on media selective for a particular "species" of the rhizobiaceae, such as S. meliloti (#Barber 1979 #Bromfield 1994#Kinkle 1994),R.leguminosarum(#Louvrier 1995), R.tropici(#Soberon-Chavez 1989), bradyrhizobia (#Gault and Schwinghamer 1993 #Tong and Sadowsky 1994#Gomez 1993), Astragalus sinicus rhizobia (#Cao 1972), Agrobacterium (#Bernaerts 1963#Schroth 1965#Clark 1969#New 1971) etc.

Due to the scatteredness of literature reports, rediscoveries are the rule rather than the exception: for instance, a resistance of rhizobia to copper sulfate had been reported as early as 1907 (#Simon 1907see Mueller 1925 for -negative- review), and been rediscovered by #Tong and Sadowsky in 1994 allowing for a publication lag, probably been independently reported by #Biro in 1995and was also (then independently) found by us with the type strain of Bradyrhizobium japonicum. In a similar manner, a stimulating influence of 5-20 ppm manganese on rhizobial nitrogen fixation had been reported by #Olaru and #Rocasalano in 1915(apparently independently), then been utilized for an Agrobacterium-selective medium by #Clark (1969 inaware of the earlier publications), and again been emphasized in paper on arctic rhizobia metal resistance (#Appanna 1991). Interestingly, the manganese content in #Pagan's medium (1975), a classic medium for ex-nodulo rhizobial N fixation, is also rather high (no explanation given). In preliminary experiments, we also tried manganese as a selective agent, but were unable to achieve significant results.

The most common method for medium-termed storage of rhizobia is on agar slants, which at 4oC usually last 3-6 months. A better method for long-term storage is in 15% glycerol at -15oC or -70oC, but repeated freezing and thawing have to be avoided.

1.4.4.2. Identification of rhizobia

----------------------------------------------------------------------------------------- Table T-1.4-A: Root nodule bacteria taxonomy: state of the art (#Lindstroem 1996) ----------------------------------------------------------------------------------------- Genus Rhizobium
R. leguminosarum biovar viciae (#Jordan 1984) "vetch rhizobia"
R. leguminosarum biovar trifolii (#Jordan 1984) "clover rhizobia"
R. leguminosarum biovar phaseoli (#Jordan 1984) "French bean rhizobia"
R. loti (#Jordan 1984) "lotus rhizobia"
R. huakuii (#Chen 1991) "Astragalus sinicus rhizobia"
R. galegae (#Lindstroem 1989) "goat's rue rhizobia"
R. tianshanense (#Chen 1995proposed) "Tianshan rhizobia"
R. tropici (#Martinez-Romero 1991) "tropic rhizobia"
R. etli (#Segovia 1993) "etl rhizobia"(French bean)
R. ciceri (#Nour 1995proposed) "chickpea rhizobia"
R. hainanense (#Gao 1994 proposed) "Hainan rhizobia"

Genus Sinorhizobium
S. meliloti (#Jordan 1984) "medic rhizobia"
S. fredii (#Scholla 1984) "fast-growing soybean rhizobia"
S. saheli (#Lindstroem 1996) "Sahel rhizobia"
S. teranga (#Lindstroem 1996) "tropical tree rhizobia"

Genus Azorhizobium
A. caulinodans (#Dreyfus 1988) "stem-nodulating rhizobia"

Genus Bradyrhizobium
Bradyrhizobium japonicum (#Jordan 1984) "slow-growing soybean rhizobia"
Bradyrhizobium elkanii (cf. #Zhang 1996) "Elkan rhizobia"
Bradyrhizobium liaoningense (#Xu 1995proposed) "very slow-growing soyb. rhizob."

Genus Agrobacterium
epiphyletic group of non-nodulating bacteria within the Rhizobiaceae
------------------------------------------------------------------------------------------

In most soils there are many strains of rhizobia competing for each host plant, for example, #Dowling (1986) claims to have found as many as 42 different S. meliloti strains on 100 alfalfa plants on a 100m2 field with alfalfa plants! So, when working with rhizobia one cannot assume that any soil is really free of them, however indigenous strains might sometimes be of low productivity. This shows that, for doing any research on rhizobial inoculation techniques, it is very important to identify strains correctly. There are a variety of techniques available for this and present numerical taxonomy emphasizes that none of them is the ultimate way, however the information different approaches yield can vary. Recent advances in rhizobial taxonomy have been reviewed frequently (#Chen 1985, #Wang 1992, #Yang 1993, #Martinez-Romero 1994,#Lindstroem 1995).

Reviewing current methods in microbial taxonomy #Vandamme (1996), classified techniques by their range of data generation: For very high-order taxonomic distinctions, DNA sequencing is optimal, for a rather broad range from the family level to species/strain differentiation phenetic assays can be used, at the species-to-family range cell wall structure and fatty acid analyses give good results, for species-genus differentiation DNA-DNA hybridization and %GC content analyses can give good results and for the strain-species microlevel RFLP,DNA-amplification, serology and MLEE can give optimal results. Some of these methods will be introduced below.

1.4.4.2.1. Phenetic assays

The expression of the genotype is in intricate correlation with the environment, so that phenotypic characteristics are less stable than genetic data. On the other hand, from a practical point of view, phenetic characteristics reveal pretty much direct information on the organism in an environment (albeit artificial) and the ways to culture it. So, in most studies of totally unknown bacterial strains a phenetic characterization is still among the first steps.

1.4.4.2.1.1. Host-plant assay

Potential host plants (selected according to the different inoculation groups) are grown from surface-sterilized and then rhizobium-inoculated seeds either on agar slopes or Erlenmeyer flasks or bigger sterile or semi-sterile arrangements. After several weeks nodulation is observed.

1.4.4.2.1.2. Biochemical tests

Biochemical tests are usually quite simple observations on bacterial growth under certain limiting conditions, such as adverse pH, salt concentrations, inhibiting dyes, utilization of carbon and nitrogen resources, antibiotic resistance and more calibrated assays such as the Gram stain, acid production, melanin production etc.

1.4.4.2.1.2.1. Why there is no straightforward biochemical test for rhizobianess

Most bacterial species have initially been defined by their biochemistry, and are thus - per definition - amenable for a determination by a set of biochemical tests which categorize the strain. Usually, by this philosophy (most successfully employed by the Bergey's Manual of Systematic Bacteriology), it is easy to determine at the genus and with more labor the species of a given strain. Unfortunately, rhizobia do not fit to this definition: there are defined as organisms to form nodules on legumes, regardless of their biochemistry. So there are -to date- no easy and crucial test except the host-plant assay which however often gives false negatives.

1.4.4.2.1.2.2. Determining rhizobianess by numerical taxonomy

Due to such practical difficulties, rhizobial classification was a very silent field in the aftermath of W.W.II, but become more vivid again with the advent of numerical taxonomy (made feasible by the advent of cheap computing power). Numerical taxonomy uses similarities in phenetic or genetic data obtained by the researchers to cluster different entities into different taxa (see #Sneath and Sokal (1973 Chinese edition: 1984) for in-depth discussion). For example, by the application of numerical taxonomy, #Graham (1964) showed that Agrobacterium and Rhizobium are closely related (Rhizobium being a paraphyletic group), this was confirmed by DNA-DNA hybridizations (#Heberlein 1967). Of course, to make this data reliable, regardless whether in botany (#Stacey 1980) or microbiology, it is recommended to apply at least 50-100 characteristics. Characters that are frequently included are carbon and nitrogen source utilization, pH ranges, salt, antibiotics and dye resistance, growth speed, colony morphology, growth temperature ranges, biochemical tests etc. (#Parke and Ornston 1984#Chen 1988#Zhang et al.1991#Sun et al.1993#Gao 1994#Novikova 1994#Chen 1995).

The most common method for assessing similarity between two strains is the simple matching coefficient which counts the ratio of (shared positives + shared negatives) / number of all tests (#Sneath 1973), sometimes only (shared positives) / (shared positives + differing tests) are counted. (#Sneath 1957). However, it must be pointed out that as we are dealing with microbe culture growth, there is one important consideration: cultures may vary in their growth speed and thus very similar cultures might simply appear very different, for example:

    Test    1       2       3       4       5       6
StrainA y y y n y y
StrainB n y n n n n
StrainC n n y y y y
(y=growth,n=no growth)

On a swift glance, the similarity A/C is 50%, A/B is 33% and B/C is 17%. However, note that all differences between A and B could also simply be ascribed to the fact that A grew better than B (maybe the number of cells spread on the plates was higher etc.). #Sneath (1968) has thus developed a concept for eliminating the influences of vigor in bacterial taxonomy which will be discussed in more detail in section 4.

As far as the classifying algorithm is concerned, for numerical taxonomy there are basically two approaches, one is bottom-up, the other is top-down. When talking about phylogenetic tree the "top" is the root (with higher taxonomic hierarchies), the "bottom" are leaves and fine branches (with lower taxonomic hierarchies). The bottom-up algorithm (cluster analysis) finds out which two OTUs (operational taxonomic units, e.g. strains or clusters of strains) are most similar and clusters them together, that is subsequently treats both taxonomic units a single new one. This step is repeated until all units are clustered together. By this principles the relatedness of strains can be classified into different hierarchies and is often represented in a tree diagram.

Methods for cluster analysis mainly vary in how to generate hypothetical new OTUs from old ones and have been extensively reviewed by #Sneath 1973 In recent times, the UPGMA algorithm reviewed by #Chen (1986) is employed most widely (at least in numerical taxonomy),i.e. if a OTU1 consists of M strains and OTU2 consists of N strains then their Euclidian distance (simply the square root of (1-similarity coefficient)) is the average of the Euclidian distances of all possible comparisions (N*M).

On the other hand, the top-down algorithm is theoretically more recent and elegant, as it tries to find out a direct evolutionary most parsimonious tree. It basically assumes that a tree assuming minimal evolution is best and can proceed by setting those OTUs most different above the roots of a binary tree. Subsequent - always those most different - OTUs are thus clustered to the appropriate branches of that tree (see #Zhong 1990:162 for an example). According to this principle (always find the most different OTUs and connect them to the most similar branch of the already existing tree) one can generate a reasonable tree; note however that for the mathematically challenfing ("NP-complete") problem of finding the shortest tree most commercial software for sequence data employs far more sophisticated optimization algorithms.

From the above-mentioned principles it can be inferred that the bottom-up approach will be especially reliable for the low-order taxa, while the Wagner-Farris top-down approach rather emphasizes higher-order taxa. The bottom-up approach also has the advantage that the effect of different ways of assessing similarity can be seen quite directly right with the clustering steps while the whole tree is needed to say anything about their usefulness in a top-down arrangement.

1.4.4.2.2. Genetic assays

The classical numerical taxonomy (previous section) has a weakness that (a) the amount of labor involved is high (b) the choice of biochemical tests will inevitably lead to some distortion in the taxonomy. For example, classical numerical taxonomy had great difficulties in assigning even higher cluster. A major breakthrough was thus #Woese's (1987) paper who used slowly changing 16S rRNA sequences to propose a three kingdom system (Archaebacteria,Eubacteria and Eukaryota). We will introduce four popular approaches to obtain data on genetic diversity, viz. isoenzyme analysis, 16 S rRNA sequence analysis, length analysis of restriction enzyme generated fragments (RFLP) and amplification of repetitive sequences by the polymerase chain reaction. Again it is emphasized there is no data from phenetic or genetic analysis methods which is a priori "best", and a taxonomist should use data obtained by different methods(#Graham 1991).

(1) Isoenzymes

Strictly speaking, this method also use phenetic data, but at least it is close to "sequences", that is proteins from different organisms known to have identical functions were electrophoresized, stained with protein-specific dyes and the results were compared. Unless the proteins underwent different posttranscriptional modifications amino acid sequence changes should affect electrophoretic mobility (#Selander 1986). This method was first developed for eukaryotes in the 60s, and only in the 80s became widely used in microbiology, for rhizobia see (#Martinez-Romero 1991#Eardly 1990#Demezas 1991#Souza 1992).

(2) 16S rRNA sequence analysis

This name is a bit misleading: what is sequenced is not rRNA but rather the DNA sequence coding for it, this sequence being chosen because rRNA accounts for 80% of cellular RNA which means that it is repeated many times. So historically, it was relatively easy to obtain by reverse transcription. Today, choosing appropriate primers a fragment of the desired variability can be simply amplified by PCR, be sequenced and compared to database sequences.

So in rhizobial taxonomy, 16s rRNA sequence analysis has been used to for taxonomy for wider phylogenetic relationships. For instance, #Young (1991) used it to establish that a photosynthetic rhizobium strain BTAi lies near the B.japonicum type strain; #Xu (1995), #Nour (1995), #Segovia (1993) and #Martinez-Romero(1991) used it to establish new species (Bradyrhizobium liaoningense, Rhizobium ciceri, R. etli and R. tropici). Using 16S RNA data, #Willems and Collins (1993) and #Sawada (1993)published dendrograms which basically has confirmed the existing taxonomic data on Rhizobium and Agrobacterium. For an interesting example of applying sequence diversity to one of the host plants, the New World Astragali, see #Wojciechowski (1993).

(3) Restriction fragment length polymorphisms

In contrast to technique (2) which uses all information of a tiny fraction of the genome, techniques (3) and (4) use a fraction of information on the whole genome. In conntrast to technique (1) which analyzes mutations in exons, (3) can also detect mutations in introns around expressed exons.

The idea of restriction fragment length polymorphism is to cut the whole genome of each organism, electrophorize it and to hybridize it onto nitrocellulose filters containing to known radioactively or biotin-labeled sequences (Southern blotting). This laborious technique has been applied for rhizobia by #Kaijalainen and Lindstroem (1989) and #Eardly 1990

(4) Genetic fingerprinting by PCR

Principles of genetic fingerprinting: since the DNA-DNA hybridization experiments by #Britten and Kohne (1968), the existence of repetitive DNA in eukaryotes ("C value paradox") has been known for more than a quarter of a century. The purification of accurate DNA replication enzymes has made DNA amplification feasible, and the first paper introducing the polymerase chain reaction (PCR) to the public (#Saiki 1985) dealt with the detection of sickle cell anemia alleles in an eukaryote (man). Very soon have repetitive sequences such as Alu (in mammalia) been used for "genetic fingerprinting" (#Nelson 1989), i.e. the amplification of highly variably distributed sequences to identify individuals. While 16SRNA sequence analysis is still the standard tool for the higher-taxa bacterial taxonomist, for rather fine intrageneric or intraspecific analyses, genetic fingerprinting compares well with the more laborious isozyme analysis or serological data.

There are two common approaches: as a primer either choose any short random sequence and hope that it is adequately replicated, this approach is called RAPD, see #Williams 1990 #Welsh and McClelland 1990 #Hui Dongwei 1992#Cao Jiashu 1995#Bingen 1995#Woodburn 1995#Fadi 1995),in rhizobium it has been used by #Harrison (1992) and #Dye (1995), on the host plant medicago by #Brummer (1995).

With an increasing amount of bacterial DNA sequences available, more subtle repetitive elements have been found in bacteria, too (for rhizobia see #Flores 1987), and a second (more direct) approach became possible, that it to choose sequences which are known to be sufficiently repeated. Actually most "arbitrary" primers used by RAPD aren't arbitrary at all (and indeed it would be wasting time and resources to calibrate each experiment for truly arbitrary primers). For example, van #Belkum (1995) terms his methods "arbitrarily primed PCR", but among others, uses the ERIC2 (see below) primers. In other words, the distinction is quite blurred. The most prominent repetitive elements found in the 1980s first in E.coli, then in a wide range of bacteria, were termed REP (repetitive extragenic palindromes; #Stern 1984) and ERIC (enterobacterial repetitive intergenic consensus) sequences. Their role is still unclear, but except that it could be "selfish" DNA, it seems most probable they have some control function (#Versalovic 1991), possibly involved in DNA polymerase or DNA gyrase binding (#Dimri 1992#Lupski 1992). Versalovic (ibd.) constructed 12 outward-directed primers (18-36bp) based on these sequences, and used PCR and electrophoresis generate strain-specific fingerprints. Of these primers, he chose two pairs of primers (one pair for ERIC, one pair for REP), widely distributed in the eubacteria, which since then have been widely applied. Some authors (#Cassol 1994working on HIV) reported better results for REPs, others (#Rodriguez-Barradas 1995working on the rhizobium relative Bartonella) found ERICs yielding more information. Some only used REPs (#Georghiou 1994), others only ERIC (#Liu 1994)

As this method is very promising for the large rhizobial genomes (#Sobral 1991),rhizobium workers began to use it very early (#Bruijn 1992 #Judd 1993) on S. meliloti and Bradyrhizobium japonicum.

1.4.4.3. Assessing the effectiveness of inoculants

To date, the only reliable way of doing this is to expose plants to rhizobia. This can be done in similar arrangements used for host plant range determination This can be done either in the lab (#Gibson 1980) or in the field (#Vincent 1970), e.g. by genetically marked strains (#Wilson 1995) allowing easy assays, for a recent review of general current techniques see #Herridge 1995

1.4.4. Inoculant production

If the productivity of a given strain has been clearly established, one can ferment it on a medium scale to produce inoculants. Although pure broth can be used for field testing, most commercial inoculants use carriers increasing the longevity of the products, such as peat (#Rodriguez-Navarro 1991), polyacrylamide (#Hedge 1992), vermiculite (#Ning 1993), soil powder (#Liu 1991) and talc powder (#Wang 1989). When producing inoculants, viable rhizobia should outpass the number of seeds by three to five magnitudes, depending on seed size (#Olsen 1994) and plant species (#Patrick 1995).

1.4.5. A word on basic research such as physiology and genetics of rhizobia The intricate details described in section 1.1.2.3. have been elucidated by many physiological and genetic methods, and the rhizobiaceae are probably among the 20 best-researched families of bacteria. #Brockwell (1995) gives a review on the potential of engineered rhizobia applications. Genetic engineering of rhizobia can lead to very fast breakthroughs (such as the successful transfer of a S. fredii 3.7kb enhancement factor to Bradyrhizobium japonicum, #Zhang 1994), but not every apparent progress will work out that well in the field. For example, supernodulating mutants have been created which form more nodules than normal strains, however these supernodulating mutants often cause more harm then benefit when applied in the field, as they force plants to an unnatural nodulation behavior (#Buttery 1992).

Other plants have been manipulated to bear nodules by the herbicide 2,4-dinitrophenol (#Nie 1991), but at the moment it is doubtful whether the yield responses reported by these techniques (see for example #Liu 1993) have indeed been caused by rhizobial N fixation. For example, some brassicaceae are by inheritance susceptible to broad-host-range bradyrhizobia which indeed elicit nodules on these plants. However, as the brassicacean root tissue is poisonous to the bacteria, no infection occurs, and all N-fixation effects can be attributed to azospirilla accumulating outside the nodule (#Brockwell 1995).Thus, as far non-leguminous N fixation is concerned, in the moment it might be more reasonable to have a closer look at the asymbiotic root fauna to achieve direct results.

So although these are very promising approaches which are of paramount importance for our understanding of this symbiosis, other symbioses and plant pathogens (#Wilson 1995), however when embarking on research projects in this area, one should not always expect very fast breakthroughs for direct applications.

1.4.6. Legume improvement

One should not forget that rhizobia only make up for half of the symbiosis, and there is good potential for legume breeding, for reviews see #Herridge 1995 (general) and #Ranalli 1995 (European pulses).

1.5. Aims of this work

As relatively little work has been done on Qinghai-Tibet plateau rhizobia, this thesis aimed to lay a foundation for that field: this includes the observation of naturally occurring legumes and to investigate their nodulation status and the soil nutrient situation at Xiaman. A preliminary inoculation trial with vetch rhizobia was included. It is hoped that by the identification of naturally occurring rhizobia as well as by proposing media for identification more directed inoculation should become possible.