Potential for nitrogen fixation in the roots of Pennisetum purpureum

SUMMARY The potential for nitrogen fixation in the rhizosphere of Pennisetum purpureum was studied. Preliminary investi gations showed that above-ground and below-ground bio mass productions were 4.58±3.84 and 1.6S±0.60 kg m· l. Analysis of roots/rhizomes for carbon substrates showed that the levels of starch and organic acids were very low. Differences also occurred in the values between roots and rhizomes. The levels of reducing sugars and miscellaneous soluble sugars were considerably higher. Rhizosphere soil under stands of this grass hadpH of 5.25-5.4 and was found to be sandy loam with high humus content. Highly significant correlation was found to exist between biomass production and total numbers of potential Nl-fixing bac teria (r "" 0.98) and between counts of Azospirillum sp. and titratable acidity. In contrast, significantly very low correlation existed between reducing sugar content of roots/rhizome/! and total anaerobic popUlation. This was also observed between aerobic bacteria and miscellaneous soluble carbohydrates. The high biomass production ob served could partly be attributed to the Nz"fixing potential of associated Azospirillum sp. Potentiel pour laftxation de !'azote dans les racines de Pennisetum purpureum. Le potentiel pour la fixation de I'azote dans la rhizophfre de Pennisetum purpureum a ete etudie. Des etudes preliminaires montraient que les productions de la biomasse en sur-sol et sous -sol etaient 4.58 :1:3.84 et 1.675 :1:0.60 kg/m,2, respectivement. Des analyses des substances carbonlques de racincs/rhi zomes montraient que les niveaux des amidons et des acides organiques etaient tres bas. y a eu aussi des differences entre les racines et les rhizomes. Les niveaux de sucres reducteurs et d'autres sucres solubles etaient plus hauts. Le sol de rhizophere qui supporte eet herb a eu un pH de et etait Iimon-sableux avec un taux de "humus.


Introduction
In nature, biological fixation of atmospheric nitrogen is widespread among prokaryotes. This process is catalysed by the enzyme nitrogenase. There exists no conclusive evidence of nitrogen fixation by eukaryotes (Dalton, 1974). There has been a great deal of interest in the study of free-living nitrogen-fixing micro-organisms (Mulder, 1975). correlation negative tIes importante existait entre la teneur en sucre reducteur des racines/rhizomes et la population totale anaerohique. Cctte correlation negative a aussi ete observee entre la bacterie aerobique et des diverses hydrates du carbone soluble. L'accroisement de la production de la biomasse observe peut tltre attribue partiellement au potentiel de fixer Nt par l' Azospirillum spp. production has been reviewed by Stewart (1975) and Dobereiner (1 977).
Much of the work done in the tropics all over the world in respect ofNz-fixation by roots of;grasses centred on determining or conf111l1ingthe Nt·flXing potential of tropical grasses, and showing how plant, micro-organism oredaphic factors affect this fIXation (Mishustin & Shilnikova, 1971). The present study, along with a series of others in other grasses, seeks to show the relationship between biomass production, carbon-substrate levels in roots/rhizomes and densities of potential N2 -fIXing bacteria.

Sites, samples and sampling
Samples of the grass studied were all collected at s*s along the Choba-Port Harcourt Road, the University of Science and Technology Campus and the East-West Road up to Eleme in the Rivers State, Nigeria. These locations are within a20-km radius of the city of Port Harcourt.
The above-ground and below-ground tissues of the plant were obtained for studies on biomass production while the carbon substrate content was determined on the below-ground tissues. Potential nitrogen-fixing bacterial populations were determined on the below-ground tissues. Rhizosphere soil samples were taken for the determination of physico-chemical characteristics.
Samples were collected from within 50 cm 2 quadrats laid out within the grass vegetation. For biomass production studies, plant height and numbers of shoots within the quadrat were determined after which the shoots were excised at the soil surface level. The entire root/rhizome system from within each quadrat was dug up and shaken loose of adhering soil. Soil clumps were further carefully examined for fme roots which were also collected. Below-ground tissues and rhizosphere soils were stored in plastic bags and analysed within 1-2h after collection.
Sampling was commenced duringmid-November and completed by the beginning of April.

Sample analysis
Biomass production. For this parameter, both the above-ground and below-ground tissues at each site were analysed by proximate measurements using a beam balance. Pre-drying treatments included cutting samples into shorter lengths for easy handling. These were spread on a wire gauze in smaller portions and dried at 90 °C for several days in a drying oven (B and T Model, Searl Company Limited, England) with intermittent weighing until constant weights were achieved. The fmal weight taken after drying represented biomass production calculated as kilogram per square metre (kgm· 2 ). The difference between the fresh and dry weights represented the water content of the tissues.
Determination of carbon-source content in under-ground tissues. The carbon substrates determined were reducing sugars, miscellaneous soluble carbohydrates, organic acids as titratable acidity and starch.
For the quantitative determination of reducing sugars, the Hagedarn-Jensen method, which is based on the quantitative oxidation by potassium ferricyanide, was adopted as outlined by Allen et al.{l974).
Misc~lIaneous soluble carbohydrates were de~ termined by proximate analysis using a spectrophotometer(Hitachi Model, SEMAC, Japan). The method adopted was the anthrone procedure out~ lined by Allen et al. (1974).
Forthe determination of organic acids, only the water-soluble acids were considered. THratabJe acidity was determined as described by Milton & Waters (1955).
Starch was extracted with perchloric acid and determined bytbe colorimetric modification procedure involving formation of a blue complex with iodine as outlined by Allen et al. (I 974).

Soil analysis
In the soil analysis, the following were determined: SoilpH, organic nitrogen, inorganic nitrogen as ammonium nitrogen (NH/-N) and nitrite nitrogen (N02'-N).
Soil pH was determined by the soil-in-water method as described by Black (1965) using a pH meter (Model PYE UNICAM, Philips, England).
Organic nitrogen was determined by ad igestion system using the traditional Kjeldahl procedure as outlined by Allenetal. (1974).
Inorganic nitrogen: NH 4 + -N was determined in rhizosphere soil samples by the procedure outlined by Allen et al.(1974). Absorbance was measured at 635 mm using a spectrophotometer (Hitachi ModeJ, SEMAC, Japan). Soil extract for NO l ' N analysis was obtained in the same manner as de~ scribedforNH 4 +-N by Allen etal. (1974). Forthe determination ofNO; -N in the extract, the method of Barnes & Folkhard (1951) was used. Absorbance was read at 543 mm in using a spectrophotometer(Hitachi Model, SEMAC, Japan).

Microbiological analysis
Roots/rhizomes collected during the sampling exercise were shaken vigorously to remove adheringsoilafterwhichtheywerewashedwithrunning tap water to remove all residual soil. Afterwashing, samples were left in a tray for about J 0-20 min for water to drain. Twenty gramme portions, made up of a 1: 1 ratio of roots and rhizomes, were weighed and then surface-sterilized in 1 per cent chloramine-T solution for 1 h. After sterilization, tissues were then rinsed with sterile distilled water 3 times, by soaking them in the water for 10 min. at a time to remove all trace of the Chloramine-T. Surfacesterilized tissues were asepticaJlymacerated using sterile mortar and pestle. The macerated tissues were then transferred aseptically into conical flasks containing 180 ml sterile distilled water to give a lin-l0 dilution from which subsequent dilutions were made.
Enumeration of potential N2 -fixers such as aerobes, microaerophiles, strict and facultative anaerobes was done using the MPN count method (Abd-EIMaJek, 1971). Fortheestimationofaerobic and microaerophilic N2 fixers, growth with the formation of surface and subsurface pellicles, respectively, were regarded as positive. In addition to the microaerophilic and aerobic Nz-fixers, facultative anaerobes were also enumerated in aerobic cultures since gas bubble formation indicative of nitrogen-fixing strict/facultative anaerobes was observed. Anaerobic growth was indicated by turbidity and gas formation in cultures. For the spirilla, actual microscopic examination of all aerobic tubes was carried out to ascertain the presence of these micro-organisms, since pellicle fonnation had always been associated with the occurrence of Spirillum sp. In addition, microscopic examination of the anaerobically-incubated tubes was also carried outto ascertain the possibi thy of spiri Ila thriving in anaerobic cultures.
Medium B was used for the cultivation of strict and facultative anaerobes. It was essentially a modification of Medium A, exceptthat it contained no agar, but contained 0.2 g sodium thioglycate and O.lg ascorbic acid added to the autoclaved portion before sterilization to enhance anaerobiosis. Sterilization and dispensing were done as for Medium A in rubber cap-stoppered culture test tubes. After inoculation, the surface of the medium in the tube was layered with a sterile plug of agar (2%) to further enhance anaerobiosis. All cultures were incubated at 37°C.
NumberofN 2 -fixing bacteria was computed by the MPN technique described by Abd-EI-Malek (1971 Group A (production of acid and gas from glucose-peptone water) -MRVP, indole and urease production, lactose fermentation, citrate utilization, starch hydrolysis, oxidase test and detection of catalase.
Group B (production of acid only from glucosepeptone water) -Same as for Group A.
Group C (no change in glucose-peptone water) -requirement of biotin, malate and mannito I utilization, and hydrolysis of starch.
Cultural and morphological tests were performed for each group and included colony characters such as pigment production on solid agar plates, slime production, motility and cell shape. For Groups A and B, biochemical, cultural and morphological tests were essentially those outline by Cruickshank et al. (1975). For Group C, a modified semi-solid version of Medium A was used. Three sets of this modified medium were prepared, each set having malate, mannitol or starch as sole carbon source.

Biomass production
The range of biomas production of different stands of Pennisetum purpureum is presented in Table 1. Shoot biomass production ranged between 0.74 and 8.42 kg m,2 dry weight with a mean value"of4.58±3.84kgm· 2 • Similarly, production in roots/rhizomes ranged between 1.05 and2.25kg m,l dry weight with a mean value of 1.65 ± 0.60 kg m,2. Shoot height, number and water content are also presented in Table 1.

Soluble carbon substrate content in belowground tissues
Results of studies involving the determination of levels of different soluble carbon substrates in below-ground tissues of the plant are outlined in Table 2. Data presented show that marked differences occurred in the concentrations of these carbon sources. Reducing sugars were of higher concentrations (18.13 ± 5.39 in roots and 46.98 ± A comparison of concentrations of carbon substrates showed that appreciably wide differences existed when roots and rhizomes are considered except for starch and organic acids. Higher con-

Physico-chemical characteristics of rhizosphere soil
Analyses ofrhizosphere soil samples revealed that soils under stands of plants of Pennisetum purpureum were generally sandy loam or sandy loam with high humus content with pH values ranging between 5.25 and5.4(5.34±0.05).
Variations were observed in the nitrogen contents of rhizosphere soil samples. The mean or-ganicnitrogenlevelwas2.97± L47mg g.l. Similar variations also existed between inorgan ic nitrogen levels as shown in Table 3. Nitrite-nitrogen (NCT 2 -N) levels were much higher than those of ammonium-nitrogen (NH+ 4 -N).

Microbiological analyses
Counts of aerobic, microaerophilic and stricti facultative potential N 2 -fixing bacteria associated with rhizosphere of P. purpureum are presented in Counts of strictlfacuItative anaerobes (gas producers in anaerobic culture) showed a mean value ofl2.04±7.93 x 106cel\sg-1 dryweight (Table 4).
Slightly lower counts were recorded for facultative amier9bes (gas producers in anaerobic cultures). Microscopic examination of anaerobic cultures showed the presence of spirilla and constituted 2.5 per cent of anaerobic cultures.
A common feature of Azospirillum sp. and related organisms is that being microaerophilic, their growth is characterized by the formation of a pellicle beneath the surface of semi-solid nutrient medium in tubes. This characteristic growth is often used to estimate the numbers of these organisms in cultures. The actual presence of these organisms was confirmed in this study by microscopic examination of tubes showing pellicle formation. A comparison of counts obtained by the above~mentioned criteria showed that by pellicle formation only, Azospirillum sp. and related organisms were as high as 24.0x 10 6 cells per gram of tissue. By contrast, no counts were recorded when actual microscopic examination of such cultures was carried out. Similarly, a culture with a count of 54.0 x 10 6 cells by subsurface pellicle formation had by microscopic examination, 0.68 x 10 6 of Azospirillum sp. Morphological, physiological and biochemical tests of isolates from all cultures showing surface and subsurface pellicles, gas production in aerobic and anaerobic cultures showed that Enterobacter sp. constituted 50 per cent, Pseudomonas, 12.5 per cent and Azospirillum and related organisms, 12.5 per cent oftotal isolates. Other unidentified organisms also constituted 12.5 per cent. Organisms of the genus Klebsiella (Aerobacter) were conspicuously absent as presented in Table 5.

Statistical analysis
Statistical analysis was carried out to determine possible negative or positive correlation between different parameters studied. Fig. 1  there was a very high positive correlation between the number of potential N2 ~fixing bacteria in the rhizosphere of Pennisetum purpureum and the amount of total dry matter produced by this plant (r = 0.98). An identical trend was obtained when concentrations of organic acids (as titratable acidity) in roots/rhizomes are compared with the corresponding counts of Azospirillum sp. in aerobic cultures, in (Fig. 2.) At a concentration of 0.125 mg g'] titratable acidity, the bacterial count was zero. A similar attempt at demonstrating possible correlation shows that a very low positive correlation (r=0.34) which was not significant at the 95 percent confidence limit, existed between the concentrations of reducing sugars in roots/rhizomes and counts of total anaerobic bacteria. Fig. 3 shows that at a concentration of24 mg g'l reducing sugar counts were 16.0x10 6 and 160 x l0 6 cfu/g. A count of 160 x 10 6 cells was also recorded at a concentration of 47 mg g,l.
In order to investigate the existence of any correlation between the concentration of readily available and oxidizable carbon substrates in the roots/rhizomes, statistical analyses were carried out Results showed that there was the absence of any significant correlation between numbers of anaerobic bacteria and concentrations of reducing sugars(r=O.34; Fig.3). Sbnilarly, Fig.4showsthat there was the near absence of correlation between numbers of aerobic bacteria and concentrations of miscellaneous soluble carbohydrates (r = 0.03).
Discussion aad conclusion A number of bacteria have been found to be associated with nitrogen fIXation in the roots of some tropical grasses including Panicum maximum, Pennisetumpurpureum andAndropogon tectorum (Dobereiner & Day, 1976). The bacteria include some members of the Enterobacteriaceae notably Enterobacter cloacae and Klebsiella pneumoniae which were conspicuously absent in the grass studied, Azospirillum sp., the aerobic Bacillus spp., and the obligately anaerobic clostridia (Knowles, 1975).
In grasses there is the current belief that the major N1-fixing bacteria belong to the genus Azospirillum (Bulow & Dobereiner, 1975). Results obtained by comparing the formation of pellicles in semi-solid media and the actual occurrence of spirilla from microscopic examination of culture tubes, indicate thatthe number of Azospirillum sp.
may in factbe small compared to that of the aerobic and anaerobic N 2 -fixers. The high numbers obtained by previous workers, for example Bulow & Dobereiner (1975), Dobereiner & Day (1976) and Balandreauet al. (1975), may in fact have been due to their associating pellicle formation with the occurrence of spirilla in the roots of Pennisetum purpureum and similar grasses. That this may have been the case can be seen more clearly in Table 4 where numbers of spirilla attributable to counts based on observation of pellicle were quite high as compared to the low numbers obtained based on actual microscopic examination. Based on this fmding, it is suggested that actual microscopic examination is necessary to positively determine the oocurrence of Azospirillum in counts of N zfixers.
In some grasses, such as wheat, other grassbacteria associations have been found. Neal & Larson (1976) described a very specific association of one wheat line with a Bacillus sp. and a lower incidence of total bacteria in the rhizosphere. In the present ~tudy, the results obtained showed greater numberofaerobic N1-flxers compared to the spirilla. Similarly,numbers of the facultative/strict anaerobicNz"fIxers, including Enterohacter, were relatively high. In a similar study, Balandreau et al. (1975) have shown that numbers of aerobic N z fixers were higher than those of the anaerobic species in the rice plant. A comparison of counts presented in Table 4 shows a striking relationship between horizontal parameters under aerobic and anaerobic considerations.
The amount of nitrogen fixed by plants would in general be determined partly by the number ofN 2fixing heterotrophic bacteria on their roots. This number in tum is determined by other factors including the levels and availability of carbon substrates, soil pH and levels of inorganic ions (Paul, Meyer & Rice, 1971;Mulder & Brotonegoro, 1974). The amount of nitrogen thus fixed would determine biomass production since nitrogen is essential for the build-up of tissue proteins. Fig.l shows more or less linear correlation (r-=O.98; P=Q.05) between total numbers of potential Nz-fixers and total dry matter production.
Assuming the nutrient conversion rate of a plant is low, it could be speculated thatthere would be a build-up of absorbed nutrients in the plant tissues to such an extent that they become actually inhibitory to the plant's development. The nutrient conversion rate would be indirectly related to the amount of root system developed. In plants with a high nutrient conversion rate and high biomass production, there would be the need to develop a corresponding efficient root system where shoot and root/rhizome dry matter production levels are positively correlated (Adoki, 1984).
The numbers as well as development of potential N 2 -fixing bacteria on grass roots are determined to great extent by the presence and availability of carbon compounds (Dobereiner, Day & Dart, 1972) and very low amounts of combined nitrogen (Macura & Kunc, 1961) so that wide CfN ratios would favour N z -fixation (Jensen, 1965;Abd-E1-Malek,1971). In the grass studied, wideCfNratios were observed between levels of total soluble carbohydrates, reducing sugars and NH/-N and N0 2 '-N. The types of carbon substrates available determine the predominant Nz-fixing group (Brouzes, Mayfield & Knowles, 1971;Paul, Meyers & Rice, 1971). Thatthis is the case can be seen in the association of Azospirillum with organic acids where most strains do not use sugars but commonly grow on malate (Dobereiner & Day, 1976). Generally, low levels of organic acids and high levels oftota) soluble carbohydrates and reducing sugars occurred in the roots/rhizomes of this grass. The latter forms, not effectively utilized by the Azospirillum group, could be responsible for their rather low numbers in the grass.
lthas been observed (Dobereiner & Day, 1976) that although Azospirillum /ipoferum is quite ca~ pable of fixing N 2 , it can grow more rapidly when supplied with a source of combined nitrogen, espe~ cially ammonia (Burris, Okon & Albrecht, 1976). Table 3 shows a very low level of soil NH / ~ N, which could also have contributed to the low numbers observed for the Azospirillum group. If similar conditions exist in soils under maize culti~ vation, this result could partially explain the observed low level ofN input in maize roots associated with Azospirillum by (Burris, 1976;Burris, Albrecht&Okon, 1977).
On the other hand, these findings may be explained if the levels of these carbon substrates are determined seasonally, for example in the temperate grass Spartina where these have been reported to show a seasonal pattern corresponding to that offIxation rates (patriquin & McClung, 1981). Table  2 shows a relatively high level of reducing sugars but less of starch. Starch is a reserve or storage carbohydrate, therefore, its level would be drastically reduced in grasses that are still in active growth. Since the grass studied has a seasonal pattern of growth, the levels of storage carbohydrates would show a corresponding variation. Starch, like otherreserve carbohydrates, would be hydrolysed for use during the growing season. This could partially explain the very low levels of starch, since samples were collected when the grass was still in active growth.