Influence of growth phases and desiccation on the degrees of unsaturation of fatty acids and the survival rates of rhizobia


Dr Patrice Mary, Institut Universitaire de Technologie ‘A’, Département Génie Biologique, Bd Paul Langevin, Cité Scientifique, BP 179, F-59653 Villeneuve d’Ascq Cedex France (e-mail:


The influence of growth phase on the evolution of cellular fatty acids (CFA) and survival of Sinorhizobium and Bradyrhizobium during desiccation and storage at different levels of relative humidity (R.H.) was studied. Lactobacillic, cis vaccenic and palmitic acids were the major fatty acids of S. meliloti RCR 2011, B. elkanii USDA 120 and B. japonicum 3.2, whatever the growth phase. An exchange of cis vaccenic with lactobacillic acid was observed during the course of growth. The degree of unsaturation (% unsaturated CFA/% saturated CFA = u/s ratio) was significantly higher during the mid logarithmic phase of growth. Survival rates immediately after desiccation were unaffected by the growth phase and the R.H. Furthermore, no correlation was found between survival rate and u/s ratio. During the course of desiccation, the u/s ratio of rhizobia decreased but the decrease was largely independent of the R.H. Optimum R.H. values for storage were in the range 22–67·8%, and S. meliloti was significantly more tolerant than the bradyrhizobia. Cells of rhizobia harvested in the lag phase of growth were more resistant to protracted storage than cells at other growth phases. Again, no correlation was found between u/s ratio and survival rates, despite the expected practical significance for screening for drought-tolerant micro-organisms.


As rhizobia are able to fix atmospheric nitrogen in symbiosis with legumes, they represent a way towards environmentally sustainable agriculture. Rhizobia are thus economically important not only because of increased agricultural legume yields, but also for the enrichment of the soil with nitrogen for other crops, and for the protection of soil and water from pollution caused by the use of chemical fertilizers.

Desiccation, salinity and temperature are substantial and often lethal stresses for rhizobia in the soil and during the preparation and storage of inoculants ( Smith 1992). The survival of rhizobia during desiccation and protracted storage is influenced by the presence and nature of protective media or solutes ( Cliquet & Catroux 1994), the rate of drying ( Mary et al. 1985 ; Fouilleux et al. 1994 ), high initial cell concentration before drying, relative humidity (R.H.) values, the temperature and nature of the atmosphere during storage in the dried state ( Mary et al. 1993 ; Paul et al. 1993 ), and rehydration ( Salema et al. 1982 ; Kosanke et al. 1992 ).

The physiological status of cells influences the response of bacteria to a number of stresses. It is generally assumed that non-growing cells are more tolerant of desiccation than their actively-growing counterparts ( Bale et al. 1993 ; Castro et al. 1995 ). However, the effect of growth stage on the tolerance of bacteria to desiccation stresses and storage in the dried state is unclear as both no effect ( Dye 1982; Caesar & Burr 1991; Cliquet & Catroux 1994), and a greater tolerance of stationary phase cells ( Amarger et al. 1972 ; Mary et al. 1986 ; Teixeira et al. 1995a ), have been reported.

Many authors consider the outer membrane and/or the cytoplasmic membrane to be the principal site/s of the lethal damage caused by desiccation ( Salema et al. 1982 ; Teixeira et al. 1995a ; Teixeira et al. 1995b ; Castro et al. 1996 ; Teixeira et al. 1996 ; Castro et al. 1997 ). During the slow decreases or increases in water potential, the increased viability of bacteria can be related to a slow water flow across the cell membrane which avoids alterations of the membranes ( Kosanke et al. 1992 ).

It is known that alterations in the environmental conditions induce changes in the fatty acid composition of microbial membrane lipids ( Sajbidor 1997). The fluidity of the lipid bi-layer is thus affected, and these changes probably enable vital membrane functions to continue. Desiccation stresses ( Zikmanis et al. 1982 ; Kieft et al. 1994 ; Castro et al. 1996 ; Teixeira et al. 1996 ), associated stresses such as decreases in water activity ( Hilge-Rotmann & Rehm 1991; Heipieper et al. 1996 ), increases in temperature ( Suutari & Laakso 1992; Heipieper et al. 1996 ; Théberge et al. 1996 ) and the growth phase ( Mackenzie et al. 1979 ; Hubac et al. 1992 ; Suutari & Laakso 1992; Drici-Cachon et al. 1996 ) affect, particularly, the fatty acid composition of bacterial lipids.

The fatty acid profiles of bradyrhizobia and rhizobia are quite different and have been used for chemotaxonomic purposes ( Jarvis & Tighe 1994; Graham et al. 1995 ). Apart from the effects of growth phase ( Mackenzie et al. 1979 ; Hubac et al. 1992 ) and of different temperatures ( Théberge et al. 1996 ), no information is available on the consequences of other environmental stresses on the fatty acid profiles of these organisms.

The aims of the current study were to evaluate the impact of the growth stage of cells of rhizobia on their survival and fatty acid composition during desiccation and storage at different R.H. levels.

Materials and methods

Bacterial strains

Sinorhizobium meliloti RCR 2011 (Rothamsted Collection of Rhizobium, Harpenden, UK), Bradyrhizobium elkanii USDA 120 (United States Department of Agriculture, Beltsville, MD) and Bradyrhizobium japonicum 3.2 (kindly provided by Dr C. Bonnier, Gembloux, Belgium) were used in this study.

Medium and buffer

All strains were grown in Yeast Extract Mannitol (YEM) broth of the following composition (g l−1 distilled water): K2HPO4, 0·5; MgSO4.7H2O, 0·2; yeast extract, 1; NaCl 0·1; Mannitol, 10. The pH was adjusted to 7·0. A buffer containing 1 g K2HPO4 and 0·2 g MgSO4.7H2O l−1 distilled water was used to wash the cells and prepare a 10-fold dilution series.

Desiccation experiments

Sinorhizobium and Bradyrhizobium strains were harvested from YEM broth at the lag, the middle of the logarithmic and the stationary phase of growth. Cells were washed twice by centrifugation (7500 g for 20 min at 4 °C) and suspended in buffer. About 2 × 109 cells were placed on a sterile membrane filter (Millipore GS; 0·22 μm pore size; 25 mm diameter) by rapid suction of 1 ml of the cell suspension. The filters were then placed on dry absorbent pads to remove the excess water. Samples were dried and stored in small desiccators (125 ml) in which the R.H. was controlled by silica gel (3% R.H.) or by saturated CH3COOK.1·5H2O, K2CO3.2H2O, KI and KCl solutions to give R.H. values (at 30 °C) of 22, 43·6, 67·8 and 83·5%, respectively. Slow drying occurred at atmospheric pressure as previously described ( Mary et al. 1994 ). The surviving cells were counted after the drying step and during storage. Filter samples were transferred in 2 ml of buffer, then mixed on a vortex mixer (Labo-Moderne 37, rue Dambasle 75015, Paris, France) for 1 min. Viable cells were counted by spreading 10-fold dilutions on YEM agar plates.

Fatty acid extraction and analysis

Extraction of bacterial lipids and preparation of fatty acid methyl esters (FAME) were carried out according to Miller & Berger (1985). Briefly, cells were collected from different growth phases by centrifugation (7500 g for 20 min at 4 °C). Cell pellets were washed twice, then mixed with 1 ml NaOH in methanol–distilled water solution (3 : 10 : 10, w/v/v), heated at 100 °C for 30 min and cooled at room temperature. Methylation was carried out with 2 ml methanol–HCl 6 mol l−1 solution (13 : 11, v/v) by heating at 80 °C ± 1 °C for 10 min. After rapid cooling in an ice bath, FAME were extracted during 10 min with 1·25 ml methyl tertiary butyl ether–hexane (1 : 1, v/v) and washed with 3 ml NaOH 0·33 mol l−1. The organic phase (0·8 ml) was transferred to a 2 ml Silyler vial evaporated under nitrogen flow and then adjusted to 10 μl with hexane.

Extracts were analysed by gas chromatography after injection of 1 μl onto a Shimadzu 17 A equipped with a flame ionization detector and a Shimadzu Class CR10 injector software (Shimadzu Corporation 3, Kanda-Nishikicho, 1-chome, Chiyoda-Ku, Tokyo 101, Japan). The following operating conditions were used: capillary column (Quadrex, OV-225, 30 m × 0·32 mm × 1 μm); carrier gas: hydrogen at 40 kPa; injection temperature: 220 °C; detection temperature: 240 °C; oven temperature: initially 40 °C, rising to 195 °C at 25 °C min−1, maintained for 15 min, then rising to the final temperature of 220 °C at 5 °C min−1 and maintained for 20 min. Peaks were identified by comparing their retention times with those of authentic standards (Supelco ref. 47080 and 47885; Supelco France 20 Quater, rue Schnapper 78101, StGermain-en-laye, Cedex, France).


Changes in cellular fatty acid composition of rhizobia during growth in YEM at 30 °C

Cultures of rhizobia grown to the lag, mid-logarithmic and stationary phases of growth were used for the determination of total cellular fatty acid (CFA) composition ( Table 1).

Table 1.  Changes in cellular fatty acid (CFA) composition of Sinorhizobium meliloti RCR 2011, Bradyrhizobium elkanii USDA 120 and B. japonicum 3·2 during the course of growth in YEM at 30 °C
% CFA * and u/s ratio of CFA for bradyrhizobia and rhizobia cells at different growth stages
S. meliloti RCR 2011B. elkanii USDA 120B. japonicum 3·2
  1. *% CFA, mean (±s. d.) of three separate samples. †u/s ratio, mean (±s. d.) of % unsaturated CFA/% saturated CFA; for each strain means in the same row not followed by the same letter are significantly different (P < 0·05). ‡Lag, lag phase; midlog., middle of the logarithmic phase of growth; stat., stationary growth phase. §ND, not detected. ¶T, Trace indicates less than 0·20%.

C12:0ND §T NDTT0·92 (1·05)NDT0·48 (0·44)
C14:0TTTTT0·83 (0·46)NDT0·23 (0·20)
C15:0NDTNDNDTTNDT0·18 (0·15)
C14:0-3OH5·85 (0·36)5·75 (0·42)5·28 (0·33)NDNDNDNDNDND
C16:1w7c0·91 (0·21)1·00 (0·07)0·69 (0·03)0·74 (0·24)0·54 (0·18)2·18 (2·70)T0·67 (0·16)0·78 (0·29)
C16:07·19 (0·83)5·49 (0·86)8·63 (0·35)14·39 (0·68)13·62 (0·32)16·84 (1·11)17·76 (0·40)12·39 (0·78)16·70 (0·52)
C17:0cycl.1·05 (0·28)0·45 (0·14)1·06 (0·23)T0·48 (0·42)1·06 (1·49)NDNDND
C17:00·39 (0·05)0·27 (0·01)0·56 (0·02)NDNDNDNDT0·32 (0·07)
C16:0-3OH0·57 (0·49)0·37 (0·32)0·41 (0·36)NDNDNDNDNDND
C18:1w9c0·82 (0·67)0·48 (0·27)0·59 (0·79)0·61 (0·99)ND1·23 (1·07)NDND0·77 (0·67)
C18:1w7c57·13 (4·91)72·32 (0·20)55·54 (1·18)65·12 (5·82)71·59 (5·56)56·42 (15·03)66·81 (2·96)79·15 (1·70)69·60 (1·94)
C18:02·53 (0·73)3·00 (0·16)2·71 (0·42)1·10 (0·24)1·04 (0·31)1·44 (0·57)3·96 (0·45)1·92 (0·62)2·93 (0·42)
C19:0cycl.19·73 (4·85)7·27 (2·12)20·72 (2·39)16·62 (6·72)12·37 (5·75)16·66 (8·08)9·50 (1·45)2·27 (1·02)6·53 (0·46)
C19:0-10CH30·52 (0·29)0·60 (0·08)0·44 (0·25)NDNDNDNDNDND
C20:2w6,9c0·72 (0·12)T0·60 (0·09)TTTNDNDND
C20:3w6,9,12c2·55 (0·43)2·62 (0·22)2.45 (0·37)NDNDNDNDNDND
u/s ratio4·87 (0·22)5·46 (0·50)4·53 (0·09)5·37 (0·20)5·80 (0·07)3·89 (0·42)3·52 (0·21)5·73 (0·11)3·76 (0·13)

Cells of rhizobia at both growth stages contained cis vaccenic acid (C18 : 1w7c), palmitic acid (C16 : 0) and lactobacillic acid (C19 : 0cyclo) as major CFAs (85–95% of total CFA detected). Palmitoleic acid (C16 : 1w7c), oleic acid (C18 : 1w9c), stearic acid (C18 : 0), margaric acid (C17 : 0) and cis-9,10-methylenehexadecanoic acid (C17 : 0cyclo) were also detected in small amounts (0·2–2% each). 3-Hydroxymyristic acid (C14 : 03-OH), 3-hydroxypalmitic acid (C16 : 03-OH), 6,9-all cis eicosadienoic acid (C20 : 2w6,9c) and 6,9,12-all cis-eicosatrienoic acid (C20 : 3w6,9,12c) were only observed in S. meliloti. Traces (< 0·2%) of C20 : 2w6,9c were also found in B. elkanii. Three other CFAs were intermittently observed and always in trace amounts (C12 : 0, C14 : 0 and C15 : 0).

Among the CFAs detected, the most significant differences observed during the course of growth were the amounts of C19 : 0cyclo, C18 : 1w7c and C16 : 0. The proportions of C19 : 0cyclo and C16 : 0 were significantly lower in the logarithmic phase of growth than during the other phases. On the other hand, percentages of C18 : 1w7c were higher in this same phase. The sum of the percentages of C18 : 1w7c and C19 : 0cyclo remained constant whatever the growth phase studied. The relative proportions of the other CFAs were virtually unaffected by the growth phase.

Except for B. elkanii USDA 120, the degree of unsaturation (u/s ratio) was not significantly different when cells were harvested during the lag and stationary phases of growth. The degree of unsaturation was higher in the rhizobia cells grown to their logarithmic phase.

Effect of growth phase on survival rates of rhizobia cells immediately after desiccation at different R.H. values

Survival in fractions of S. meliloti RCR 2011 and B. elkanii USDA 120 observed immediately after desiccation at five R.H. values were unaffected by the age of the cultures ( Table 2). However, for B. japonicum 3.2, survival was statistically higher for cells harvested in the middle of the logarithmic phase of growth. At this stage of growth, strain 3.2 was completely surrounded by copious amounts of capsular polysaccharides; further incubation resulted in the production of an incomplete capsule in a polar insertion.

Table 2.  Effect of culture age on the survival of Sinorhizobium meliloti RCR 2011, Bradyrhizobium elkanii USDA 120 and B. japonicum 3·2 immediately after desiccation at five relative humidity (R.H.) values
Log10 no. of cells and surviving fraction immediately after desiccation at % R.H.

and growth
phases *

initial no.
of cells
  • *

    Lag, lag phase; midlog., middle of the logarithmic phase of growth; stat., stationary growth phase.

  • †Mean (±s. d.) of two separate samples.

  • ‡Surviving fraction (N/N 0): N number of surviving cells, N0 cell number before desiccation.

  • §For a given strain and a given growth phase, Log 10 N0 of cells from the same row followed by the same letters are not statistically different at P = 0·05.

  • For a given strain, surviving fraction values from the same column followed by the same letters are not statistically different at P = 0·05.

S. meliloti RCR 2011
Lag9·37 (0·12) A §9·03 (0·21) A9·04 (0·31) A9·11 (0·12) A9·08 (0·32) A9·38 (0·08) A
10·46 A 0·46 A0·55 A0·51 A1·02 A
Midlog.9·45 (0·10) A8·80 (0·21) A8·97 (0·16) A8·96 (0·06) A9·08 (0·32) A9·38 (0·08) A
10·22 A0·33 A0·32 A0·43 A0·85 A
Stat.9·31 (0·02) A8·93 (0·19) B9·11 (0·04) AB8·93 (0·06) B9·25 (0·06) AB9·11 (0·06) AB
10·42 A0·63 A0·42 A0·87 A0·63 A
B. elkanii USDA 120
Lag9·18 (0·03) A9·01 (0·14) A9·11 (0·01) A9·03 (0·20) A9·20 (0·08) A9·25 (0·06) A
10·68 A0·85 A0·71 A1·05 A1·17 A
Midlog.9·44 (0·01) A9·47 (0·01) A9·49 (0·04) A9·33 (0·07) A9·54 (0·03) A0·40 (0·04) A
11·07 A1·12 A0·78 A1·26 A1·12 A
Stat.9·28 (0·01) A9·09 (0·28) A8·78 (0·76) A9·22 (0·13) A9·15 (0·05) A8·74 (0·37) A
10·65 A0·31 A0·87 A0·74 B0·29 A
B. japonicum 3·2
Lag8·60 (0·19) A7·70 (0·06) B7·69 (0·12) B7·75 (0·04) B8·40 (0·19) A8·54 (0·05) A
10·13 B0·12 B0·14 C0·63 A0·87 B
Midlog.9·03 (0·07) A9·11 (0·06) A8·97 (0·08) A9·03 (0·07) A9·17 (0·30) A9·22 (0·04) A
11·20 A0·87 A1 A1·38 A1·55 A
Stat.9·17 (0·04) A8·59 (0·17) AB8·15 (0·49) B8·65 (0·06) AB8·67 (0·13) AB8·41 (0·04) AB
10·26 B0·09 B0·30 B0·31 A0·17 C

Whatever the R.H. values imposed, desiccation did not significantly affect the viability of cells, except for B. japonicum 3.2 harvested during the lag phase and subjected to drying under 3, 22 and 43·6% R.H. ( Table 2). It is noteworthy that the initial cell concentration of this strain was significantly lower for cells harvested in the lag phase than in the two other phases of growth.

Evolution of the degree of unsaturation immediately after desiccation

The effects of the desiccation step at five R.H. values on the u/s fatty acid ratio of rhizobia cells harvested at different stages of growth are shown in Fig. 1. Whatever the R.H. and stage of growth studied, the u/s ratio of S. meliloti cells decreased to a threshold value of approximately 3–3·2 during the course of desiccation ( Fig. 1a). Owing to a significantly higher u/s ratio of non-dehydrated cells harvested from the middle logarithmic phase, the decreases in these indices due to desiccation were more pronounced for cells in this growth phase than for the others.

Figure 1.

Degrees of unsaturation before (To) and immediately after desiccation at 3, 22, 43·6, 67·8 and 83·5% relative humidity for Sinorhizobium meliloti RCR 2011 (a), Bradyrhizobium elkanii USDA 120 (b) and B. japonicum 3·2 (c). Cells of each strain were harvested from three stages of growth. (□), Lag; (▪), middle of logarithmic; (bsl00007), stationary phase of growth. Vertical bars indicate Standard Deviation from the mean (n = 3)

No significant decrease in the degree of unsaturation was observed after the desiccation steps for B. elkanii cells harvested in the stationary phase of growth ( Fig. 1b). This strain was thus apparently unable to decrease its u/s ratio below a threshold value of 3·6–3·8. However, cells from lag and midlog phases showed a significant decrease in their degree of unsaturation principally at R.H. values ≤43·6%.

Midlog phase cells of B. japonicum showed no reduction of u/s ratio after drying ( Fig. 1c). Cells from the lag and stationary phases reduced their u/s ratio during desiccation except when dried at 83·5% R.H.

Thus, with a few exceptions, the degree of unsaturation of desiccated rhizobia was about 30–50% lower than that of their non-dehydrated counterparts. Furthermore, the decrease of u/s ratio for a given strain and growth phase was generally independent of the level of desiccation achieved.

Effects of growth phase on the survival of rhizobia during long-term storage at different R.H. values

Whatever the species and the growth phase studied, two stages and, more infrequently, linear decreases in cell numbers were observed. Sinorhizobium meliloti showed biphasic concave survival curves at R.H. ≤ 67·8% and linear decreases at 83·5% R.H. Bradyrhizobia underwent biphasic concave decreases at R.H. ≤ 43·6% and biphasic convex or linear decreases at R.H. ≤ 67·8% (data not shown).

In order to compare tolerance to storage in the dried state among the species and growth phases studied, maximal storage times (expressed in days) for fulfilling French and Canadian inoculant quality standards were determined from survival curves ( Table 3). The standard ≥105 viable cells per seed allowed a direct comparison of desiccation tolerance of bradyrhizobia and rhizobia during storage. The ranking of susceptibility to storage in the dried state was in the order: B. japonicum > B. elkanii > S. meliloti.

Table 3.  Effects of stages of growth (lag, middle of logarithmic and stationary phases) and storage relative humidity (R.H.) (3, 22, 43·6, 67·8 and 83·5%) on the maximal storage time of Bradyrhizobium elkanii USDA 120, B. japonicum 3·2 and Sinorhizobium meliloti 2011 cells in orderto fulfil to the French and Canadianinoculants standards
Maximal storage time (days) to fulfil:
Standard ≥106 for bradyrhizobia * and ≥103 for sinorhizobia with cells harvested fromStandard ≥105§ with cells harvested from
Rhizobia species and
storage at % R.H.
  • *French standard for bradyrhizobia according to Smith (1992).

  • †French and Canadian standards for sinorhizobia according to Smith (1992).

  • Lag, lag phase; midlog., middle of the logarithmic phase of growth; stat., stationary growth phase.

  • §Canadian standard for bradyrhizobia according to Smith (1992). Maximal storage time to fulfil this standard is reported for S. meliloti in order to compare desiccation tolerance between sinorhizobia and bradyrhizobia.

B. elkanii
B. japonicum
S. meliloti

Whatever the growth phase studied, the optimum R.H. for storage was 43·6% for B. elkanii, whereas the optimum R.H. values were within the range 43·6–67·8% for B. japonicum and 22–43·6% for S. meliloti according to the standard retained or the growth phase studied.

Within the range of optimum R.H. for storage, lag phase cells of S. meliloti and B. elkanii, followed by stationary phase cells, survived better than midlog cells. For B. japonicum, a ranking of stationary > lag > midlog phase cells was observed.


The profiles of CFA observed for rhizobia are in accordance with those observed in other studies ( Jarvis & Tighe 1994; Graham et al. 1995 ). The major growth phase response of these micro-organisms in fatty acid composition was limited to an interchange of C18 : 1w7c in C19 : 0cyclo and a variation in the percentage of C16 : 0. This observation is in agreement with previous work ( Mackenzie et al. 1979 ; Hubac et al. 1992 ). Nevertheless, the evolution of hydroxy fatty acids during the course of growth, especially for rhizobia, has not been reported in the literature. The present results show that this class of CFA remain constant during growth. The interchange of oleic acid with dihydrosterulic and lactobacillic acids during the cell cycle of Gram-positive species such as Lactobacillus fermentum ( Suutari & Laakso 1992) and Leuconostoc oenos (Drici-Cachon et al. 1996) has also been reported.

Cyclopropane fatty acids are synthesized preferentially as bacteria enter the stationary phase of growth, and it has been suggested that they might protect bacteria from adverse environmental conditions such as high temperature and low pH ( Suutari & Laakso 1992; Drici-Cachon et al. 1996 ). However, Pseudomonas aureofaciens is highly susceptible to moderate desiccation (94% R.H.), although a significant increase in the ratio of cyclopropane fatty acids to their monoenoic precursors is observed ( Kieft et al. 1994 ). Cyclopropane fatty acids are generally counted with unsaturated CFAs because of their similar physical properties, particularly with regard to membrane fluidity ( Suutari & Laakso 1992; Teixeira et al. 1996 ). Thus, the evolution of degree of unsaturation during the cell cycle of rhizobia observed here was mainly a consequence of changes in the percentage of palmitic acid.

An inverse relationship between the viability of desiccated yeast and the degree of unsaturation of CFAs of non-desiccated counterparts has been demonstrated ( Zikmanis et al. 1982 ). Owing to the lack of a deleterious effect of desiccation on survival rates of rhizobia, such a correlation was not observed in the present study, although the u/s ratio was significantly higher when cells were harvested in the middle of the logarithmic phase of growth.

The absence of deleterious effects of desiccation at different R.H. on survival rates of rhizobia is in agreement with previously published work ( Dye 1982; Mary et al. 1994 ). Dye (1982) reported that growth stage had no effect on survival of rhizobia immediately after desiccation at different R.H. levels. In contrast, young cells (24 h) of Sinorhizobium meliloti (formerly Rhizobium meliloti) are more tolerant of freeze-drying than older (96 h) cultures ( Amarger et al. 1972 ). However, in the Gram-positive Lactobacillus bulgaricus, cells at stationary phase are more resistant to spray-drying than cells in the exponential phase of growth ( Teixeira et al. 1995a ). The susceptibility of B. japonicum 3.2 harvested in the lag phase was probably a consequence of a lower initial cell concentration. Indeed, it has been demonstrated that higher initial cell concentrations lead to higher survival rates during storage in the dried state ( Mary et al. 1993 ; Fu & Etzel 1995).

During the course of drying, the rhizobia decreased their degrees of CFA unsaturation and these decreases were largely independent of the R.H. values. Decreases in the u/s ratio have already been described for aw ( Hilge-Rotmann & Rehm 1991), moderate desiccation to 94% R.H. ( Kieft et al. 1994 ) and spray- or freeze-drying stresses ( Castro et al. 1995 ; Teixeira et al. 1996 ). The results reported here have extended these findings to the genera Sinorhizobium and Bradyrhizobium and covered a greater range of R.H. values. The magnitude of the decrease in u/s index due to desiccation described here was also in agreement with other work ( Castro et al. 1995 ; Teixeira et al. 1996 ). However, a more pronounced decrease at high R.H. (59%) compared with low R.H. (0 and 11%) has been reported ( Castro et al. 1995 ). The lack of adaptation in cells of B. japonicum 3.2 harvested in the midlog phase of growth could be explained by the synthesis of capsular polysaccharides, which is an energetically expensive process. This may prevent further CFA adaptation, particularly in the oligotrophic conditions encountered during desiccation.

In the absence of protective substances, rhizobia were found to be susceptible to storage at all R.H. levels and particularly at 0 and ≥67·8% R.H. The optimum R.H. values for storage of rhizobia, described here, were in accordance with previous reports ( Mary et al. 1993 ; Paul et al. 1993 ; Castro et al. 1995 ).

In general, higher survival rates during storage were observed when rhizobia were harvested in lag phase compared with the two other phases of growth. As far as is known, the highest resistance of lag phase bacterial cells so far reported was for tolerance of Escherichia coli and Enterococcus faecalis to a 60 °C heat-shock stress ( Hansen & Riemann 1963). Although the interpretation of this phenomenon is important from the practical standpoint of inoculant technology, its mechanism is unknown and further research is needed.

Although the determination of u/s ratios may be a convenient and rapid (6 h) tool for screening for drought-tolerant micro-organisms, as demonstrated by Zikmanis et al. (1982) for Saccharomyces cerevisiae, such a correlation was not found for rhizobia at different growth stages. However, it may be of practical significance for inoculant technology to adjust the fatty acid composition of rhizobia by manipulation of the medium, pH, temperature and/or aw in order to optimize the production of cells tolerant to desiccation.