Microb Ecol (2011) 62:257–264 DOI 10.1007/s00248-011-9845-4
NOTES AND SHORT COMMUNICATIONS
Bacterial Communities Associated with Chenopodium album and Stellaria media Seeds from Arable Soils Leonard S. van Overbeek & Angelinus C. Franke & Els H. M. Nijhuis & Roel M. W. Groeneveld & Ulisses Nunes da Rocha & Lambertus A. P. Lotz
Received: 21 January 2011 / Accepted: 2 March 2011 / Published online: 19 March 2011 # Springer Science+Business Media, LLC 2011
Abstract The bacterial community compositions in Chenopodium album and Stellaria media seeds recovered from soil (soil weed seedbank), from bulk soil, and from seeds harvested from plants grown in the same soils were compared. It was hypothesized that bacterial communities in soil weed seedbanks are distinct from the ones present in bulk soils. For that purpose, bacterial polymerase chain reaction denaturing gradient gel electrophoresis (PCR– DGGE) fingerprints, made from DNA extracts of different soils and seed fractions, were analyzed by principal component analysis. Bacterial fingerprints from C. album and S. media seeds differed from each other and from soil. Further, it revealed that bacterial fingerprints from soilrecovered and plant-harvested seeds from the same species clustered together. Hence, it was concluded that microbial communities associated with seeds in soil mostly originated from the mother plant and not from soil. In addition, the results indicated that the presence of a weed seedbank in arable soils can increase soil microbial diversity. Thus, a change in species composition or size of the soil weed seedbank, for instance, as a result of a change in crop management, could affect soil microbial diversity. The consequence of increased diversity is yet unknown, but by virtue of identification of dominant bands in PCR–DGGE fingerprints as Lysobacter oryzae (among four other species), it became clear that bacteria potentially antagonizing phytopathogens dominate in C. album seeds in soil. The role of these potential antagonists on weed and crop plant growth was discussed. L. S. van Overbeek (*) : A. C. Franke : E. H. M. Nijhuis : R. M. W. Groeneveld : U. N. da Rocha : L. A. P. Lotz Plant Research International BV, Wageningen University and Research Center, Droevendaalsesteeg 1, 6708 PB Wageningen, The Netherlands e-mail:
[email protected]
Introduction The trend in arable farming to move towards the implementation of integrated or organic farming systems with a reduced reliance on herbicides often coincides with an increased size of soil weed seedbanks (e.g., [1, 20]). Enhancing the role of field margins as refugia for biodiversity may also lead to more diverse weed communities within fields, and thus to an increased or a more diverse weed seedbank. If microbial communities associated with soil weed seedbanks are distinct from communities present in bulk soil, then it seems logical to assume that microbial diversity in soils will expand when the soil weed seedbank increases in size or becomes more diverse. Largersized soil weed seedbanks in arable soils are obviously a menace for agricultural production because of the weed plants emerging from these seeds compete with the crop for resources. In addition, weed seeds can sustain populations of phytopathogens that attack crops, as reviewed in Franke et al. [8]. Seeds may be critical for the survival of some pathogens, especially when arable fields are without vegetation during certain periods of the year, e.g., in wintertime in temperate climates. On the other hand, weed seeds may also harbor functional microbial groups that protect plants against phytopathogens, the so-called antagonists. This particular aspect hardly has been explored in scientific literature and may be a challenging new area in soil microbiology. Little is known about the structure and functioning of microbial communities associated with soil weed seedbanks. The presence of seeds can help bacteria to persist in soils over extended periods of time. Bacteria were even found to be present in seeds that were frozen for thousands of years in permafrost in Siberia [24]. It was speculated that weed plants can be a source of plant growth-promoting rhizobacteria belonging to the groups of Firmicutes,
258
Pseudomonas, and Stenotrophomonas maltophilia, among several other bacterial groups [26]. It is unknown whether these bacteria are transmitted via seeds to emerging weed plants or that soil-borne bacteria are stimulated in their growth near the roots of emerging weed plants. A clear distinction between microbial communities associated with soil weed seedbanks and bulk soils was never made before [8]. Hence, it is uncertain which soil microorganisms are typically associated with weed seeds. Weed seed-associated microbial populations in soil generally were ignored in the past. Discrimination between microbial communities associated with weed seeds and bulk soils is a first step for investigating microbial populations typically related with soil weed seedbanks. Different microbial techniques can be applied to investigate microbial communities in seeds. Those depending on microbial cultivation may be less useful because the vast majority of microbial species from soil and other ecosystems are poorly cultivable [25]. Most likely, poorly cultivable bacteria also will be present in weed seeds in soils, and therefore, methods aimed to detect microbial communities in seeds are mostly based on cultivation-independent (DNAbased) approaches [7, 12]. On the other hand, novel cultivation-based techniques that are aimed to increase recovery of poorly cultivable species from soils can be explored [5], and this can be a valuable tool for recovery of bacteria from weed seeds. In a later stage, these isolates can be tested for their roles in soil functioning. In the current study, we hypothesize that bacterial communities associated with soil weed seedbanks are distinguishable from the ones present in bulk soils and more alike the communities present in the mother plant. For that purpose, seeds of weeds commonly occurring in arable cropping systems in the Netherlands were collected from soils of two different organically farmed fields. To test the assumption that the microbial communities in the soil weed seedbank originate from respective mother plants and not from soil itself, we compared bacterial communities in weed seeds collected from soil with those harvested from weed plants originating from the same area and grown in the same soil. Bacterial fingerprints (polymerase chain reaction denaturing gradient gel electrophoresis (PCR– DGGE)) made from DNA extracts from different seeds from soil and from plants were compared with each other and with the ones derived from bulk soil DNA extracts.
Material and Methods Description of Soils and Soil Collection Soils were obtained from two fields from the same organic leek production farm located near the village of Mierlo, The
L. S. van Overbeek et al.
Netherlands (N 51.452, E 5.603). The soils of both fields were characterized as “sandy,” and soil of field 1 had an organic matter fraction of 2.6% and a pH of 5.6 and that of field 2 had a slightly higher organic matter fraction (2.9%) and lower pH (5.3). Both fields were treated according to standards for organic farming. Fields under organic farming management were chosen to avoid any impact of soilapplied biocides on microbial communities. Soils of the two Mierlo fields were organically enriched by the same manure application, i.e., 25 tons/ha pig manure and 30 tons/ ha cattle manure per field was applied 6 weeks before leek planting. The history of crop production was somewhat different between the two fields: iceberg lettuce, endive, and leek were the crop plants in the rotation scheme of field 1, whereas it were fennel, endive, and leek that were in the crop rotation scheme of field 2. Fields were surveyed for the presence of weed species before soil sampling. At sampling (in April 2008), the leek crop in field 1 was already harvested (5 days before sampling), whereas the leek crop was still standing in field 2. For sampling, three locations of 3×3 m in size were randomly chosen over the fields. Per location, four samples were taken at 25-cm depth (with a soil borer, 2.5 cm in diameter) and pooled, leading to a total sample of approximately 1.5 kg per location. These pooled soil samples were used for later seed collections and soil DNA extractions, whereas 25 kg of soil was taken from the same six locations for later growth of weed plants collected from the two fields. For analyses, soils, seeds from soils, and seeds from plants were compared per location (n=3) over the two fields. Weed Seed Collections from Soils and Plants Soil samples (1 kg) were suspended in water and sieved (mesh size, 0.5 mm). Remaining materials (mostly consisting of small pebbles, roots, and seeds) were visually inspected for the presence of weed seeds. Collected seeds, rinsed in fresh tap water and dried to air, were temporarily stored in sterile Eppendorf vials at room temperature for not longer than 2 months. Intact Chenopodium album (Fat-hen) and Stellaria media (common chickweed) plants, two from each species per location in the two fields, were collected and transferred to the greenhouse where they were planted into pots filled with 2.5 kg of soil from the same location in the field as where they were taken from. Plants were allowed to grow up to maturity in the open air with a coverage for protection against solar radiation, wind, and rain. Mature seeds were collected from plants and pooled per plant species and location in the field. Individual seeds were randomly taken from each batch for later microbial analyses.
Bacterial Communities Associated with Chenopodium album
Bacterial Recovery from Seeds Bacteria from C. album and S. media seeds were isolated on oligotrophic agar medium [22] as described in Da Rocha et al. [5, 6]. Therefore, pooled seeds (two per location) were crushed in a sterilized mortar and suspended in 1 mL of sterile water. Suspensions were tenfold serial diluted in sterile saline (0.85% NaCl in demineralized water), and 100-μl volumes per dilution were plated in duplicate. Plates were incubated at 25°C in an atmosphere containing 4% CO2 and 16% O2, and colonies were counted after 15 days. DNA extractions were made from bacteria grown on plates that had received undiluted samples (see below). Average colonyforming unit (CFU) numbers between seed fractions per species and location in the field (converted to log values) were compared by two-way ANOVA (Genstat release 12.1, Hemel Hempstaed, UK). DNA Extractions from Seeds, Soils, and Bacteria For DNA extraction from seeds, four C. album or S. media seeds per replicate sample were bead beaten with four chrome steel beads (2.38 mm in diameter) in sterile 2.0-mL screw cap tubes (BIOplastics, Landgraaf, the Netherlands) in a bead beat apparatus (Hybaid Ribolyser, Hybaid, Middlesex, UK) set at speed level 4 and run for two times for 30 s. DNA extracts were made from bead beaten seeds using the Qiagen DNeasy Plant kit (QIA GEN, Germany), according to the procedure provided by the manufacturer. For DNA extraction from soils, 0.5-g subsamples were used, and extractions were made with the Mobio Power Soil kit (MO BIO) in accordance to the procedure provided by the manufacturer. For DNA extraction from bacteria, cells scraped from plates were suspended in sterile saline (plate wash), and DNA was extracted from these suspensions using the MasterPure™ Complete DNA and RNA Extraction Kit (Epicentre Biotechnologies, Madison, USA). All DNA extracts were visually inspected for quality and quantity in ethidium bromide-stained agarose (Invitrogen, CA, USA) gels (0.8%), run in 0.5×TRIS acetate buffer for 3 h at 30 mA. Molecular Fingerprint Analyses on Seed, Soil, and Bacterial DNA Extracts by PCR–DGGE Bacterial PCR–DGGE fingerprints were made from all DNA extracts. PCR amplifications, done on 20 ng DNA in standard 50-μL reaction mixtures [28], using primers 968 F (with GC clamp [16]) and 1378R [10, 11] were performed in a PTC-100 thermocycler (MJ Research Inc., MA). Products were checked for the expected size in ethidium
259
bromide-stained agarose gels. DGGE was done in polyacrylamide gels (6%) with a denaturing gradient of 45–65% (100% denaturant consists of 7 M urea and 40% formamide). Individual lanes in gels were loaded with a mixture of 10 μL PCR product (c. 200 ng) and 5 μL loading buffer (0.25% bromophenol blue, 0.25% xylene cyanol FF, and 30% glycerol). All gels were run in a PhorU2 apparatus (Ingeny, Goes, NL) set at 60°C and 100 V and run for 16 h. After running, gels were stained with SYBR Gold (Molecular Probes, Leiden, NL), and digitized fingerprints from individual lanes were normalized using Molecular Analyst software (version 1.61, BioRad, Veenendaal, NL). Multivariate analysis on PCR–DGGE fingerprints was performed using CANOCO 4.53 (Biometris, Wageningen University and Research Centre, NL). Band intensity and location in PCR–DGGE fingerprints were used as “species” variables, whereas nominal values for the origin of the samples (soil, seed from soil, or seed from plant) were used as “environmental” variables. Indirect gradient analysis was done by correspondence analysis (CA) and direct gradient analysis by principal component analysis (PCA). Average number of bands and Shannon diversity (H’) values (based on band number and intensity of bands) were compared between samples by two-way ANOVA. Average values were considered to be different at levels of P≤0.05. For identification of individual bands in PCR–DGGE fingerprints, DNA from selected individual bands was extracted from the DGGE gel using QIAquick PCR purification kit (Qiagen) and cloned into pGEM-T Easy Vector (Promega, WI, USA). Cell lysates from individual white colonies (containing inactivated lacZ gene) in 0.1 mL of sterile water were made by boiling for 5 min. Sequencing was done in reaction mixtures containing 5 μl of cell lysate, 1 μl DETT Dye (Dyenamic ET Terminator Cycle Sequencing Kit, Healthcare, GE), 3 μl of dilution buffer, and 1 μl (0.5 μM) of primer 1492R. Linear amplifications were performed for 25 cycles at 94°C for 20 s, 50°C for 15 s, and 60°C for 60 s. The amplified products, approximating 350 bp in size, were sequenced in an ABI prism automatic sequencer by making use of the services of Greenomics (Plant Research International, Wageningen, The Netherlands). Sequence data were first trimmed for removal of flanking vector sequences and then checked for presence of chimeras using the Check Chimera tool (http://www.ncbi.nlm.nih.gov). Nonchimeric sequences were assessed for similarity, using the Sim_ Identity index, with sequences of type and non-type strains in the Ribosomal Database Project, using the bioinformatic analysis tools of the Sapelo Island Microbial Observatory (last accession August 31, 2010; http://simo. marsci.uga.edu/).
260
L. S. van Overbeek et al.
absent in all DNA fingerprints from soil. The chloroplast band was excluded from seed fingerprint analysis.
Results Seed Collection from Soils and Weed Plants
Soil and C. album C. album, S. media, Persicaria maculosa (Redshank), Fallopia convolvulus (black bindweed), and Urtica urens (small nettle) were the weed plant species commonly observed in both fields over the past 3 years. Seeds of C. album were highest in number upon recovery from soils of both fields, 55 (15, 19, and 21 in the three locations) in field 1 and 24 (7, 7, and 10) in field 2, followed by those of S. media, 13 (5, 8, and 0) in field 1 and 7 (6, 1, and 0) in field 2. Seeds of the other plant species were lower in number or were only found in one of the two fields (Table 1). Because C. album and S. media seeds were found in both fields in sufficient quantities, these seeds were selected for further microbial analysis using cultivationdependent and cultivation-independent approaches. Bacterial Analysis of Soil Weed Seedbank The number of S. media seeds from four locations was too low (minimum of two are needed) for bacterial isolation, whereas C. album seed numbers were sufficient in all six locations over the two fields. Average bacterial CFU numbers, expressed per seed and converted to log values, in C. album seeds were 5.22 in field 1 and 4.88 in field 2, and in S. media seeds 4.99 (n=1) in field 1 and 4.27 (n=1) in field 2. These numbers were not significantly different between fields (C. album only), indicating that bacterial numbers were about the same in C. album seeds in both fields. Bacterial PCR–DGGE Fingerprint Analysis of Soil and Seed DNA Extracts Bacterial PCR–DGGE fingerprints made from all seed DNA extracts revealed the presence of a chloroplast band in the lower part of the gel (high in denaturing gradient), which was
Table 1 Identity and number of seeds from soils of different field locations
Location and field Mierlo Field 1
Field 2
SeedsThe number of bands in bacterial PCR–DGGE fingerprints differed depending on the origin of the DNA extract. Among numerous bands low in intensity, between 22 and 35 clear bands in PCR–DGGE fingerprints from field 1 and 2 soils were found, and differences between fields were not significant. From C. album seeds, band numbers were between 10 and 19, when from soil, and between 4 and 9, when from plants, and there was no significant difference in the number of bands in seed fingerprints from field 1 and 2 at both occasions. Over the two fields, average band numbers in soil fingerprints (24) were significantly higher than in seed fingerprints from soils (13) and on their turn, numbers in these fingerprints were higher than in those from seeds from plants (7). Shannon diversity values calculated from band number and intensity were highest in soil (2.67) than in seeds from soil (2.05) and lowest in seeds from plants (1.66), and differences were significant at all occasions. This indicates that bacterial species diversity differs between soil and seeds, but also between seeds from soil and from plants. S. media Because of the limited number of available S. media seeds recovered from soil (Table 1), only two samples were analyzed from soil 1 and one from soil 2. Bacterial PCR– DGGE fingerprints made of DNA extracts from S. media seeds recovered from soils (between 9 and 13 bands, n=3) and plants (between 6 and 13, n=6) were significantly different from the ones made of soil DNA extracts. Average Shannon diversity values calculated from fingerprints from soil were significantly higher (2.61) than from seeds from soil (1.46) and from plants (1.47).
Plant species name
Number of seeds
Chenopodium album (fat-hen) Stellaria media (common chickweed) Persicaria maculosa (Redshank) Fallopia convolvulus (black bindweed) Urtica urens (small nettle) Chenopodium album (fat-hen) Urtica urens (small nettle) Stellaria media (common chickweed) Persicaria maculosa (redshank)
55 13 15 4 1 24 14 7 1
Bacterial Communities Associated with Chenopodium album
261
Principal Component Analysis on Soil and Seed Bacterial Community Fingerprints A biplot made by PCA on bacterial PCR–DGGE fingerprints from soil and C. album seeds from soil and plants revealed the presence of three distinguishable clusters (Fig. 1a). The first cluster consisting of seven fingerprints was from soil (six) and from seeds recovered from field 2 soil (one). The second cluster consisted of nine fingerprints, all from seeds, five from seeds from soil and four from seeds from plants. The third cluster consisted of two samples, and both were from seeds from plants. Separation of the largest two clusters occurred along the first axis, which explained highest variation (60.0%). The second axis separated the two clusters consisting of seed fingerprints,
a 1.2
Chenopodium album
PCA axis 2 (16.0 %)
2 seed plant 2 seed plant
2
2 2 1
2 seed soil 2 seed plant
2 seed soil 1
1 seed soil 2 seed soil
1
1 seed plant 1 seed soil
1 seed plant
-0.6
1 seed plant 1 seed soil
PCA axis 1 (60.0%)
-1.0
-1.5
0.6
b
-0.4 PCA axis 2 (3.5%)
Figure 1 Biplot diagrams calculated by principal component analysis (PCA) on bacterial PCR–DGGE community fingerprints in seeds and soils. DNA extracts were made from Chenopodium album (a) and Stellaria media (b) seeds, either recovered from soil (seed soil; hatched symbols) or harvested from plants (seed plant; open symbols) or from soils (1 or 2, closed symbols) from field 1 (1, square) or from field 2 (2, circle). Explained percentages of the variation are indicated near each axis
and this axis explained 16.0% of all variation. The strongest separation was thus between fingerprints from soil and from seeds, with the exception of one fingerprint from a pool of seeds recovered from soil 2. A smaller separation was present between fingerprints from seeds harvested from plants and from soils. The PCA biplot of bacterial PCR–DGGE fingerprints from soil and S. media seeds from soil and plants revealed the presence of two distinguishable clusters (Fig. 1b), one consisting of all six soil fingerprints and the other consisting of all nine seed (from soil and plants) fingerprints. Both clusters were separated along the first axis, explaining 88.7% of all variation. There was no distinction between fingerprints from soil-recovered and plantharvested seeds.
Stellaria media 2
1 seed plant 1 seed plant 1 seed plant 1 seed soil 2 seed plant 2 seed plant 1 seed soil
2 2 2 seed plant
2 seed soil 1
1 1
-1.5
PCA axis 1 (88.7%)
1.5
262
L. S. van Overbeek et al.
Bacterial Identities in the C. album Fraction of the Soil Weed Seedbank A total of eight dominant bands in PCR–DGGE fingerprints made of DNA from C. album seeds from soil (two) and from bacterial colonies grown on oligotrophic agar medium (six) were analyzed by DNA sequencing. Fragments of between 315 and 375 bp, compared by BLASTassisted searches, revealed the presence of seven different bacterial groups. Sequences of two bands matched with sequences of species belonging to the phylum of Actinobacteria and those of the other six with sequences of species belonging to Proteobacteria, from which one was affiliated with β-Proteobacteria and the other five with γProteobacteria (Table 2). The sequences of the bands identified to belong to Actinobacteria both matched with the sequence of Thermoleophilum album type strain at low similarity levels (