The mechanisms by which seed-applied bacteria colonize the rhizosphere in the absence of percolating water are poorly understood. Without mass flow, transport of bacteria by growing roots or soil animals, particularly nematodes may be important. We used a sand-based microcosm system to investigate the ability of three species of nematodes (Caenorhabditis elegans, Acrobeloides thornei and a Cruznema sp.) to promote rhizosphere colonization by four strains of beneficial rhizobacteria. In nearly all cases, rhizosphere colonization was substantially increased by the presence of nematodes, irrespective of bacterial or nematode species. Our results suggest that nematodes are important vectors for bacteria rhizosphere colonization in the absence of percolating water.
The use of seed-applied beneficial soil bacteria as biofertilizers, biocontrol products or for bioremediation is an area of intense study [1–3]. The rhizosphere favors bacterial growth and survival, and is the area of soil where biofertilizers  and biofungicides [6,7] are targeted. It has also been suggested as a site that can enhance activity of bioremediating bacteria  (rhizoremediation). Thus, seed application of beneficial bacteria is a key strategy for introducing bacteria to soil [7,9]. The ability of different species and strains of bacteria to colonize the rhizosphere under different soil moisture conditions varies greatly . Under conditions of mass flow, transport of bacteria by water is the main mechanism of colonization for seed-applied bacteria [2,11]. In soils at or close to field capacity, bacterial motility in response to chemotaxis is thought to be important for movement into root channels or into contact with transporting agents, although motility is not generally considered a method of colonization in itself . Under drier conditions, root colonization is likely to involve either adherence to the growing root tip and mucigel utilization, adherence to extending fungal mycelia, or movement by soil animals.
It is well known that earthworms can transport bacteria [12–15]. Furthermore, it is known that they can aid root colonization by bacteria . However, there are few published studies on the influence of nematodes in root colonization. Nematodes are invariably the most abundant soil animals in agricultural soils, and may occur in extremely high numbers (>100 g−1 soil) [17,18]. While nematodes move much shorter distances than earthworms, their ubiquitous nature means that bacteria, introduced on a coated seed, are much more likely to come into contact with nematodes . The presence of nematodes in the rhizosphere has been shown to increase bacterial growth, stimulated by the products of incomplete digestion released by nematodes accompanied by increased nitrogen mineralization [20,21]. In addition, presence of nematodes in model soils enhances bacterial activity by distributing bacterial colonies over the organic substrate . There has been some work demonstrating that nematodes can act as vectors of plant pathogenic bacteria  or rhizobium [24,25]. These studies found that the nematodes distributed cells more evenly over the root surface and thus benefited the plants, and it is therefore surprising that this area of research has not been further developed.
In order to investigate the potential for nematodes to act as vectors for improved colonization of root surfaces by seed-applied bacteria, we undertook a simple sand-based microcosm study using a combination of motile Gram-negative and Gram-positive bacteria, with or without free-living bacterial-feeding nematodes of the families Rhabditae and Cephalobidae. The bacteria selected had been previously studied in rhizosphere colonization experiments and were known to remain culturable. We used colonization of the wheat rhizosphere in the absence of percolating water as a model system.
Materials and methods
Winter wheat (Triticum aestivum var. Savannah) (Advanta Seeds, UK) was selected for use in the trial as a good representative of presently used winter wheat varieties. Nematode species used were the Cephalobid Acrobeloides thornei DWF1109, a small nematode, the Rhabditid Cruznema sp. Rosario strain, a much larger nematode (both supplied by Paul De Ley, University of California, Riverside, CA, USA) and the laboratory standard Rhabditid nematode Caenorhabditis elegans N2 (Christina Lagido, University of Aberdeen, UK). Three Gram-negative pseudomonads were selected from two species. Pseudomonas corrugata 2140, isolated from Australian soil, had been reported as a potential biocontrol agent for take-all in wheat and a good rhizosphere colonizer . P. fluorescens 10585 (NCIMB, Aberdeen, UK) has demonstrated rhizosphere colonization capabilities under matric potentials of −500 kPa  and had been previously used in soil studies [4,26–30]. P. fluorescens SBW25 (CEH, Oxford, UK) is a sugar beet phytosphere isolate that has been characterized as a colonizer of roots and shoots for several plants in field and laboratory trials [31–34]. The Gram-positive isolate Bacillus subtilis MBI600 (Becker Underwood, Littlehampton, UK) is sold commercially as a biocontrol agent. All pseudomonad strains were chromosomally marked with Kanr genes. Introduced antibiotic genes were used as a selectable marker for recovery of all bacterial species. The B. subtilis was marked with the plasmid pSB340 [35,36] carrying chloramphenicol resistance. All bacterial strains were routinely cultured in Luria Bertani broth (LB) or LB agar containing the appropriate antibiotics.
Growth and preparation of nematodes
Petri dishes of Nematode Growth Medium  were inoculated with E. coli HB101 1 day prior to inoculation with the nematodes C. elegans, A. thornei and Cruznema sp. and grown at 25°C for 7 days. Nematodes were then maintained at 17°C until required, when they were transferred to a Baermann funnel apparatus  for extraction into water. Nematodes were counted using a dissecting microscope and added to microcosms at a rate of 10 g−1 sand (A. thornei and Cruznema) or 5 g−1 sand (C. elegans).
Surface sterilization, coating and germination of wheat seeds
Wheat seeds were sterilized using an oxytetracycline/silver nitrate method . For seed coating, test bacteria were grown at 25°C with shaking at 200 rpm for 20 h after which 6 ml of culture was centrifuged, washed in 1 ml of 1/4 strength Ringer's solution (Oxoid), and then re-centrifuged. The resulting bacterial pellet was suspended in sterile 1% (w/v) high viscosity Carboxy Methyl Cellulose in 1/4 strength Ringer's solution , applied to 4 g of surface-sterilized wheat seeds, air-dried for 2 h prior to being germinated on moist filter paper in the dark at 25°C for 24 h.
Preparation of microcosms
Microcosms were prepared in root trainers (60 mm deep, with a square top of 36×36 mm) containing 37.5 g sharp sand (99, 94 and 13% less than 500, 355, and 180 µm, respectively, CHAP Construction, Scotland, UK) that had been washed and autoclaved then oven-dried at 105°C. The sand was prepared at water contents of 0.104 and 0.208 g g−1 dry sand. These correspond to matric potentials of −2.6 and −2.0 kPa, respectively, as determined by the filter-paper method . Microcosms were established by adding the required amount of water (containing the appropriate numbers of nematodes) to the surface and leaving for 16 h to equilibrate before germinated seeds were planted at 1 cm depth. Microcosms were sealed in clear polythene bags and incubated in a high light Phytotron for 14 days with 12-h periods of light and dark at 13.5 and 15°C, respectively.
We did three colonization experiments. In the first, microcosms were established at 0.104 and 0.208 g g−1 dry sand, containing no nematodes, A. thornei or Cruznema sp. Wheat seeds were coated with all of four test bacteria and four germinated seeds were planted out for each investigated treatment, resulting in a total of 96 microcosms. The second experiment was similar, but included an additional nematode treatment (C. elegans) and thus comprised 128 microcosms. The third experiment used microcosms established only at 0.208 g g−1 dry sand and used all three nematode species. In addition, a dead nematode treatment was included as a further control. This consisted of a mixture of equal numbers of the three nematode species, killed by heating to 65°C for 2 h and applied to give 10 dead nematodes g−1 sand. The third experiment thus comprised a total of 80 microcosms.
Plant harvesting, root recovery and bacterial enumeration
Root systems were recovered from sand microcosms and the shoots and seeds removed. The root system was then divided into primary roots, which were measured. These were then placed into 1/4 strength Ringer's, vortexed for 5 s and left to stand for 2 h. The resulting bacterial suspension was serially diluted and four replicate 10-µl aliquots of each dilution grown on selective agar plates. Colonies were counted and the primary roots were oven-dried at 105°C for 48 h and weighed.
Nematode recovery from microcosms and investigation into nematode distribution within the rhizosphere
In the first two experiments, one microcosm of each treatment was transferred to a Baermann funnel for 16 h and extracted nematodes were then counted. In the third experiment nematodes were recovered in this manner from all microcosms.
To establish nematode distribution throughout the sand microcosm 50-ml centrifuge tubes were cut into 2-cm sections. Tubes were reassembled with masking tape, covered in aluminum foil and 50 g of sterile oven-dried sand added. A. thornei, Cruznema and C. elegans were added as previously described with all microcosms established at 0.208 g g−1 dry sand. P. fluorescens 10586 lux coated seeds were prepared, germinated and planted as previously described. Microcosms were incubated for 2 weeks in a high light Phytotron. After this time, the tubes were dismantled into sections, the sand and roots in each section weighed, and nematodes extracted using a centrifugal flotation method  and counted.
All data were log10-transformed to stabilize the variance and compared using an analysis of variance (ANOVA) General Linear Model approach (Minitab release 12.21). When ANOVA showed significant treatment effects, individual means were compared using Tukey's pairwise comparison.
Bacterial recovery from primary roots
In the first experiment, ANOVA showed significant effects of bacterial seed coat and nematode treatment, but not water content on numbers of bacterial cfu mg−1 root. Numbers of cfu mg−1 root for the various nematode treatments and interactions between bacterial seed coat and the nematode treatment are shown in Fig. 1A,B. The presence of nematodes increased bacterial colonization, and this was particularly true in microcosms inoculated with Cruznema sp. (Fig. 1A). The effects of bacterial and nematode species interacted significantly. This is largely a result of the high level of colonization of P. fluorescens 10586 in the absence of nematodes, whereas all other bacteria colonized less effectively (Fig. 1B). In addition, presence of A. thornei significantly reduced colonization by this bacterium, while this nematode significantly (P<0.05) increased colonization by P. fluorescens SBW25 and P. corrugata 2140.
In the second experiment, there were significant (P<0.05) effects for bacterial seed coat, nematode treatment and water content on colonization. Addition of all three nematode species to microcosms increased colonization at a highly significant (P<0.001) level (Fig. 1C). Unlike the first experiment, all nematode species significantly increased colonization by all bacterial seed coats (Fig. 1D). Despite the small difference in matric potential, colonization of bacteria was significantly (P<0.001) greater in microcosms that established a water content of 0.208 than 0.104 g g−1 (mean log10 cfu mg−1=2.59±0.05 S.D. and 2.28±0.05 S.D., respectively).
In the third experiment, significant (P<0.05) effects for seed coat, and nematode species were once again observed and the two factors interacted. Trends were generally similar to the second experiment. An important observation was that addition of dead nematodes did not significantly increase bacterial colonization above control levels (Fig. 1E). In this experiment, no nematode treatment significantly increased colonization of B. subtilis, and only C. elegans significantly increased colonization by P. fluorescens SBW25 (Fig. 1F).
Nematode recovery form microcosms
Nematode recovery from microcosms after 2 weeks of plant growth was low, fewer than three nematodes g−1 sand recovered in all experiments, and variable regardless of nematode species and seed coat. Survival of A. thornei tended to be greater than the other two nematode species (Fig. 2).
Nematode location in the wheat rhizosphere
Recovery of nematodes using the centrifugal flotation method again returned low counts compared with numbers added to the sand-based microcosms. Typical recovery never exceeded 5% of nematodes added for each sampled depth (Fig. 3). Numbers of Cruznema and C. elegans were lower in samples taken from the lower fractions of the microcosms, with virtually none being recovered in the 6–8-cm section. Recovered numbers of A. thornei showed a different pattern, with greatest numbers being recovered in both the 0–2- and 6–8-cm sections, and much lower numbers in intermediate depth sections.
In the model system we used, no further additions of water were made to the microcosms after establishment, preventing bacterial movement by or in percolating water. Analysis of the results indicates that all of our test bacteria were able to colonize and grow throughout a developing wheat rhizosphere, but to different extents. Differences in colonization probably resulted from differences in bacterial species ability to adhere to the root tip, utilize plant exudates and/or a lack of motility [41,42]. However, addition of live nematodes to our system substantially increased the degree of colonization for all test bacteria. When the ubiquitous nature of bacteria and nematodes in soils is considered, along with the fact that all nematodes share an environment with bacteria at one time in their life , it seems highly likely that nematodes will have a beneficial impact on rhizosphere colonization by seed-applied bacteria. The mechanisms by which nematodes increased bacterial colonization are not ascertainable from the data presented here, but probably a combination of movement of the bacteria on the nematodes cuticle , passage of viable bacteria through the nematodes digestive system  or enhancing bacterial growth as a result of nematode excretion  are probably all involved.
Differences in colonization observed between experiments might have resulted from different rates of nematode mortality. This in turn would also explain why the recovered numbers of nematodes differed at the end of each experiment (Fig. 2). Reasons for changes in mortality between experiments and treatments are not clear, but reductions in nematode numbers from the initial inoculum were recorded in every experiment possibly due to either the low pH of the washed sterile sand (4.8±0.2), or starvation. Poor nematode survival in sand microcosms has been reported previously. Decreases in nematode numbers to about 60% of the initial inoculum in sand microcosms  and in storage at pH 4–6  have been reported for entomopathogenic nematodes. Our third experiment showed that the increased bacterial colonization could not have resulted solely from increased availability of substrate and nutrients from dead nematodes. Microcosms to which dead nematode treatments had been added had levels of bacterial colonization similar to control microcosms identifying the importance of live nematodes. Whatever the reasons for the decrease in nematode numbers within the microcosms, the overall effect on bacterial colonization of adding live nematodes to the system was to increase it.
Analysis of abundance of the nematodes throughout the wheat rhizosphere indicated differences in survival rate and location of each species. Previous studies involving C. elegans demonstrated that sand in the size range of 0.125–1.25 mm did not affect nematode distribution , although in those systems, the sand was kept moist. Water is important for nematode movement, although the presence of A. thornei and not Cruznema or C. elegans in the lower fractions of the recovered sections suggests that nematode size was probably a more important factor in determining depth of nematode movement. Attraction of nematodes to active bacterial colonies has been demonstrated [18,19,45] and activity of P. fluorescens 10586 was probably highest near the root tip, where increased root exudation occurs . If Cruznema and C. elegans had been unable to locate this food source, due to either an inability to keep up with or get to the root tip, then these nematodes would likely starve and be recovered in lower numbers, as was observed to be the case. However, in trials 1 and 3 the smaller A. thornei was less successful at increasing colonization than the larger Cruznema sp. or C. elegans. These observations might imply that the larger nematodes carry more bacteria on their cuticle and increase colonization in this manner, although this was not investigated here.
Despite the observed reduction in nematode numbers during experiments, our results suggest that these animals play a useful role as vectors for bacteria colonization of the rhizosphere in the absence of percolating water.
We thank the BBSRC and the BIRE initiative for funding this work, and Mark Bailey, Tracey Timms-Wilson (CEH, Oxford, UK) and Paul DeLey (University of California, Riverside, CA, USA) for provision of cultures and technical assistance.