Assessing the impact of the biological control agent Bacillus thuringiensis on the indigenous microbial community within the pepper plant phyllosphere


  • Editor: Aharon Oren

Correspondence: Zhihui Bai, Research Centre for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China. Tel.: +86 10 62923562; fax: +86 10 62923563; e-mail:


Although biological control agents (BCAs) have been used extensively for controlling insects and pathogens of plants, little is known regarding the effects of such agents on the indigenous microbial communities within the plant phyllosphere. We assessed the effect of the BCA Bacillus thuringiensis (Bt) on the microbial communities within the pepper plant phyllosphere using culture-independent methodologies. Phospholipid fatty acid (PLFA) analysis suggested that the bacterial and fungal biomass were not significantly affected following Bt application. However, principal component analysis of PLFA data indicated that Bt did change the phyllosphere microbial community structure significantly. 16S rRNA gene-directed PCR with denaturing gradient gel electrophoresis (DGGE) also suggested a significant change in the phyllosphere bacterial community structure following Bt inoculation. Phylogenetic analysis of excised DGGE bands suggested a change in bacterial phyla; bands from untreated samples predominantly belonged to the Firmicutes, while Gammaproteobacteria abounded in the treated samples.


The aerial portion of plants provides a habitat for a diverse community of prokaryotic and eukaryotic microorganisms. The aerial habitat colonized by these microorganisms is termed a phyllosphere (Lindow & Brandl, 2003). Some phyllosphere microorganisms are plant pathogens, but most are nonpathogenic organisms that play important roles in altering plant surface properties, fixing nitrogen, promoting plant growth, controlling pathogens and degrading organic pollutants (Murty, 1984; Hirano & Upper, 2000; Krechel et al., 2002; Schreiber et al., 2005; Sandhu et al., 2007). Phyllosphere microorganisms use the limited resources available on the plant's surface. Their community composition is affected by nutrient availability and other factors such as temperature fluctuations, humidity, UV radiation and osmotic pressure, as well as the presence of foreign chemicals such as pesticides (Lindow & Leveau, 2002).

Biological control agents (BCAs) are now being used widely to control pests and plant diseases. Such agents may also alter the indigenous nontarget microorganisms within the phyllosphere and the rhizophere, thereby affecting the community's ecological and functional properties. Thus, Johansen & Olsson (2005) reported transient effects of the BCA Pseudomonas fluorescens DR54 on the barley rhizosphere microbiota, whereas P. fluorescens CHA0 did not change the diversity of culturable fungi in the cucumber rhizosphere significantly (Girlanda et al., 2001).

Bacillus thuringiensis (Bt) is one of the most widely used BCAs. It produces one or more parasporal crystal inclusions (Cry or δ-endotoxins) that are toxic to various insect species including Diptera, Lepidoptera, Coleoptera and Hymenoptera (Broderick et al., 2006). Its effects on the native plant microbial community are currently poorly understood. Using culture-based techniques, Russell et al. (1999) demonstrated that application of Bt did not affect microbial diversity on the leaf surface of Brassica oleracea. However, culture-dependent techniques are restricted by the fact that more than 90% of the indigenous microorganisms are not readily culturable using standard techniques. In recent years, culture-independent methods such as phospholipid fatty acid (PLFA) analysis and 16S rRNA gene-directed PCR–denaturing gradient gel electrophoresis (DGGE) have been used to provide information on microbial diversity and community structure. PLFA results are interpreted based on the assumption that phospholipids make up a relatively constant proportion of the cell biomass and that variations in fatty acids among taxonomic groups can be used as biomarkers so that the PLFA pattern of an environmental sample reflects the microbial community composition (White et al., 1996). PCR–DGGE separates 16S rRNA gene fragments based on differences in their resistance to chemical denaturation, generating a genetic profile of the individual bacterial species within a community (Burr et al., 2006).

The objective of this study was to assess the effects of Bt on the indigenous microbial community of pepper plant cultivars, using PLFA and PCR–DGGE techniques.

Materials and methods

Experimental design

Experiments were conducted in a greenhouse located within the Institute of Vegetables and Flowers, Chinese Academy of Agriculture Sciences, China. Pepper cultivars (Capsicum annuum L.) were planted in 160 m2 plots in a greenhouse on 10 April 2007. During the course of the experiment, the daily air temperature varied within a range of 18 and 31 °C, and the average relative air humidity was 65%. Plants were watered and fertilized in accordance with local grower practices. When the diameter of pepper fruit reached 1–2 cm, the plants were sprayed with Bt (ShangDong LuKang Biological Pesticides Co. Ltd, a wettable powder formulation containing 16 000 IU mg−1 var kurstaki active ingredient applied at 1.1 kg ha−1). Equivalent pepper cultivars without Bt addition were used as control. Triplicate leaf samples (50 g) from inoculated and control samples were collected 1, 2, 4, 6, 8 and 12 days after treatment into sterile stomacher bags, cooled to 4 °C and transported to the laboratory. PLFA analysis was conducted immediately, whereas the samples for DNA extraction and subsequent PCR–DGGE were stored frozen at −20 °C until required.

Microorganism cell extraction

Leaf samples were transferred aseptically using flame-sterilized forceps and placed in polypropylene tubes containing washing buffer (0.1 M potassium phosphate buffer, pH 7.0) and sonicated (frequency 40 kHz) for 7 min in an ultrasonic cleaning bath to dislodge microorganisms from leaves. Leaf debris was then removed by a slow-speed centrifugation step (5 min, 500 g, 4 °C) and the remaining microbial suspension was centrifuged at 7000 g for 15 min at 4 °C. Microorganisms were resuspended in washing buffer and frozen at −20 °C until processing.

PLFA analyses

Triplicate control and treated samples were used for PLFA analysis. The total lipid fractions were extracted according to Bligh & Dyer (1959). Briefly, the microbial pellets obtained from pepper cultivars leaves were transferred to glass tubes with Teflon-lined screw caps. Lipids were extracted by a one-phase chloroform, methanol and 0.15 M citrate buffer extraction solution. The lipid extract was then fractionated into neutral, glyco- and polar (phospho-) lipids by silicic acid chromatography. Phospholipids were then converted into fatty acid methyl esters (FAMEs) using a mild alkaline methanolysis reaction. An internal standard to quantify the different fractions, methyl ester of nonadecanoic acid (C19:0), was added as an internal standard. FAMEs dissolved in hexane were analyzed using a gas chromatography–mass spectroscopy (GC–MS) system (Hewlett Packard HP 6890) equipped with an HP-5 capillary column (60 m × 0.32 mm), and the mole fraction of each component was calculated.

The fatty acid nomenclature chosen for this study was described previously by Frostegård et al. (1993). Fatty acids with carbon lengths of 14 to 22 were used to analyze microbial community structure. The fatty acids i15:0, a15:0, 15:0, i16:0, i17:0, a17:0, 16:1ω7c, 16:1ω9t, 17:0, cy17:0, 18:1ω7 and cy19:0 were chosen to represent bacterial PLFAs (Frostegård & Bååth, 1996; Chinalia & Killham, 2006); a15:0, i15:0, i16:0, i17:0 and a17:0 were used as biomarkers for Gram-positive bacteria (O'Leary & Wilkinson, 1988); 16:1ω7c, 16:1ω9t, cy17:0, 18:1ω7 and cy19:0 for Gram-negative bacteria (Wilkinson, 1988; Zogg et al., 1997); and the unsaturated PLFA 18:2ω6,9, to represent fungal biomass (Thirup et al., 2003; Johansen & Olsson, 2005).

Principal component analysis (PCA) was performed to compare the PLFA profiles in the samples using spss 11.5 software (version 13.0, SPSS Inc.)

PCR–DGGE analysis of 16S rRNA gene fragments

DNA from the phyllosphere microbial community of controls and treated samples was extracted as described by Yang et al. (2001) and purified using a DNA Gel Recovery Kit (Omega Bio-Tek Inc.) following the manufacturer's instructions. The highly variable V3 region of the 16S rRNA gene was then amplified by PCR using the primer set GC-PRBA338f and PRUN518r (Øvreås et al., 1997). A GC-clamp was attached to the PRBA338f primer to enable separation of the DNA fragments on the DGGE gel. DGGE was subsequently performed using the D-Code universal mutation detection system (Bio-Rad). PCR products (25 μL) were loaded onto 10% (w/v) polyacrylamide gels containing a 20–60% linear chemical gradient (where a 100% denaturing solution contained 40% formamide and 7 M urea). Each gel underwent electrophoresis for 10 h at 60 °C with 100 V applied in 1 × TAE buffer. Gels were stained with ethidium bromide and photographed using a Fluor-S Multi Imager system (Bio-Rad). The profiles were analyzed using quantity one-4.6 software (Bio-Rad) to calculate bacterial community similarity values. Based on band presence/absence and band weighting (band density) analysis, a dendrogram was constructed using the DICE coefficient and the unweighted pair group method using arithmetic averages (UPGMA).

Sequence analyses

DGGE bands of interest were excised from the gel using a sterile scalpel, and the DNA was eluted overnight at 37 °C in sterile distilled water. Excised DNA was then reamplified using the primer pair PRBA 338f and PRUN518r as described above. PCR products were cloned into the pGEM-T Easy vector (Promega, Germany) and transformed into competent DH5α cells. The 200-bp 16S rRNA gene fragment inserts were sequenced from T7 primer sites on the vector. Sequence homology searches were performed at the National Centre for Biotechnology Information using the blast network service (blastn). Multiple alignments and distance matrix analyses were conducted using the mega 3.0 software package (Kumar et al., 2004). A phylogenetic tree was constructed using neighbor-joining analysis with 1000 replicates of bootstrap analysis. Sequence data obtained in this study were deposited in the GenBank database (Table 1).

Table 1.   DGGE band sequence identities
DGGE bandAccession numberClosest identity% identity
P1EU200333Uncultured Acinetobacter sp. (EU071491)99
P2EU409314Uncultured Proteobacterium (EU052058)99
P3EU409315Pantoea dispersa (EU100012)97
P4EU409316Uncultured Pseudomonas sp. (AM773559)99
P5EU409317Enterococcus gallinarum (EF025908)99
P6EU200325Paenibacillus polymyxa (EF634025)100
P7EU200336Uncultured Bacilli bacterium (EF706141)100
P8EU200335Enterococcus faecalis (AB244434)99
P9EU409318Pseudomonas sp. (AM110075)99
P10EU200328Clostridium sp. (AM884908)99
P11EU409319Desulfotomaculum guttoideum (AB294139)100
P12EU409320Sphingobacterium sp. (DQ530111)98

Results and discussion

PLFA analyses

Composite PLFA profiles provide an indication of microbial biomass (Ben-David et al., 2004) and give information on the community structure. Concentrations of bacterial and fungal PLFAs from the phyllosphere microbial communities did not change significantly following the addition of Bt (Fig. 1a). However, PCA analysis based on individual PLFA values demonstrated significant (P<0.05) effects of Bt on the community composition (Fig. 2). The ratio of Gram-positive bacteria (GP) to Gram-negative bacteria (GN) lipid biomarkers increased slightly 2 days after Bt application (Fig. 1b).

Figure 1.

 Response of pepper phyllosphere microorganisms during the 12 days following Bacillus thuringiensis BCA treatment. (a) Bacterial PLFA and fungal PLFA, (b) GP/GN PLFA profiles. Error bars represent the SE of the mean (n=3).

Figure 2.

 Scatter plot of the (a) score and (b) loading values of the first two principal components in a PCA of relative molar abundance of fatty acids originating from the phospholipids in the total microbial community in pepper cultivar phyllosphere. The day after treatment is shown by the individual marker. Error bar represents the SE of the mean (n=3).

The distribution of the control/treatments in the PCA score of PLFAs provides information regarding the effect of Bt application on the phyllosphere microbial community, where PC1 and PC2 account for 47% and 26% of the variation, respectively. PLFA patterns from treated plots were clearly different from those of control plots, and scores from the treatment coordinated to the right in the score plot along PC1, while the controls were located to the left (Fig. 2a). This suggests that Bt treatment altered the phyllosphere microbial community. Along PC2, the control and treated samples showed some variation, indicating that succession time (represented by PC2) was another possible factor responsible for the variation. The loading of the individual PLFAs is given in Fig. 2b. The fatty acids i14:0, a15:0, i15:0, i16:0, i17:0, 16:1ω7c, 16:1ω9t and cy17:0 were most important for the differentiation of the treated samples. The loadings showed that PC1 had a positive relationship with the saturated PLFAs i14:0, a15:0, i15:0, i16:0 and i17:0. These species are found in the right part of the loading plot, indicating that these tended to increase following the Bt treatment. On the other hand, the unsaturated PLFA, 16:1ω7c, 16:1ω9t and cyclopropane cy17:0 were negatively correlated with PC1, indicating that these PLFAs decreased under the Bt treatment. PLFA found in the midsection of PC1 were not affected by the treatment.

DGGE analysis

As shown in Fig. 3, DGGE profiles of samples following Bt treatment differed from the control samples, with the appearance of new bands (bands P2, P3, P5 and P9) and disappearance of other bands following Bt treatment (bands P1, P7 and P8). The relative intensities of band P12 increased after the treatment. DGGE analysis of the control samples revealed relatively persistent banding profiles during the course of this study. Only minor changes were observed: band P1 was detected at 1 day and subsequently decreased in relative intensity, and bands P10 and P11 only appeared at days 4 and 6 and then disappeared. Seven bands were detected in the DGGE profiles from Bt-treated samples. Bands P9 and P12 were detected in all treated samples, but their relative intensities decreased at day 12. Band P6 first appeared at day 4 and then increased in relative intensity until day 12, whereas P2, P3, P4 and P5 also appeared at day 8 but increased slightly in intensity at day 12. We did not detect Bt in the DGGE profile, indicating that Bt is not a dominant species within the phyllosphere following its application to the plant surface. Bt is probably a poor competitor for leaf surface nutrients compared with indigenous epiphytic bacteria (Maduell & Armengol, 2008).

Figure 3.

 DGGE profile for PCR-amplified fragments of the 16S rRNA gene for control (C) and treated (T) samples from 1 day to 12 days following Bacillus thuringiensis BCA treatment.

The effect of Bt treatment on the phyllosphere community was further confirmed by dendrogram analysis (Fig. 4), where two distinct clusters were observed between the Bt-treated and the control samples (similarity coefficient less than 45%). We found significant changes (P<0.01) in the microbial community composition following the introduction of Bt. Temporal changes in bacterial community structure were also detected by the grouping of profiles according to sampling dates within clusters I and II (Fig. 4). In cluster I, two smaller clusters separated the samples acording to sampling date: days 1, 8 and 12 clustered together and separately from the other treated samples. Also in cluster II, the phyllosphere bacteria communities showed some variation. Samples from 2 and 4 days produced similar profiles (similarity coefficient >90%), but the samples from 8 and 12 days clustered together (similarity coefficient >90%) separated from the earlier samples.

Figure 4.

 UPGMA cluster analysis of DICE distance matrix calculated from DGGE banding patterns (based on presence/absence and band weighting).

The results of the sequence homology searches for the 12 bands labeled in Fig. 3 are presented in Table 1. Their phylogenetic relationship with other closely related species is shown in the neighbor-joining tree in Fig. 5. The sequences belonged to members of the Firmicutes, Gammaproteobacteria and Bacteroidetes. For the control sample, 6 bands were sequenced and consisted of five members of the Firmicutes (P6, P7, P8, P10 and P11) and one member of the Gammaproteobacterium (P1). A total of six bands were identified from the Bt-treated samples: four members of the Gammaproteobacteria (P2, P3, P4 and P9), one member of the Bacteroidetes (P12) and one member of the Firmicutes (P5).

Figure 5.

 Neighbor-joining analysis showing the phylogenic relationship of 16S rRNA gene sequences from excised DGGE bands with other related organisms. Band identities are described in Table 1.

Final comments

Our study of the microbial changes in the phyllosphere of pepper plant cultivars following the application of Bt, using culture-independent techniques (PLFA, PCR–DGGE), showed significant effects of the treatment. There have been other reports of the effect of biological control agents on plant microbial communities. In addition to the above-mentioned work by Johansen & Olsson (2005) on the effect of P. fluorescens DR54 on the microbiota in the barley rhizosphere, there is a study by Pfender et al. (1996) on the effect of Limonomyces roseipellis on plant microbial community structure and function. Significant differences were not found in all cases: the two biological control agents Serratia plymuthica HRO-C48 and Streptomyces sp. HRO-71 did not affect culturable bacteria on the Verticillium host plant in strawberry field trials (Scherwinski et al., 2007), Bt treatments did not appear to affect the culturable microbial community within the phyllosphere of Brassica oleracea (Russell et al., 1999) and Bt and its crystal protein had no influence on the population sizes of culturable heterotrophic bacteria and saprophytic fungi (Ferreira et al., 2003). In our study, PLFA offered a powerful approach to demonstrate the effect of Bt application on the composition of phyllosphere microbial communities of pepper plants. Application of Bt may affect the indigenous microbial populations due to direct competition for nutrients on the plant surface, or due to the crystal protein that Bt produces and that may be used as a substrate by some microorganisms or may be toxic to others.

Future studies will investigate the influence of different biological control agents on the composition and structure of phyllosphere microbial communities in other plants. A deeper understanding of the mechanisms of the interactions between biological control agents and the indigenous microbial community might improve the efficiency of the biological treatments and also lead to enhanced yields of agricultural crops.


This study was funded by the National Natural Science Foundation of China (no. 30600082). We thank Prof. Baoju Li and the Institute of Vegetables and Flowers, Chinese Academy of Agriculture Sciences, for supplying a greenhouse and pepper plants.