THE IMPACT OF HABITAT PATCHINESS AND BACTERIAL GENOTYPE ON SPREAD OF BACTERIA IN THE PHYLLOSPHERE

THE IMPACT OF HABITAT PATCHINESS AND BACTERIAL GENOTYPE ON SPREAD OF BACTERIA IN THE PHYLLOSPHERE

Susan S. Hirano^a, Brendan K. Riely^a, Keith D. Fourrier^a, L. Stuart Baker^a, and Christen D. Upper^a,b

^aDepartment of Plant Pathology and ^bUSDA/ARS Plant Disease Resistance Research Unit, University of Wisconsin, Madison, WI 53706

SUMMARY

Our overall objective is to examine the impact of the patchiness of the agroecosystem and the variability of bacterial adaptation to the phyllosphere on the likelihood of spread of phyllosphere bacteria. Two specific objectives were addressed in field experiments: To assess the effects of the substrate over which bacteria adapted to the phyllosphere must travel on spread and the ability of bacteria to grow in the phyllosphere on the likelihood of spread. Key findings included the following: Bacterial spread can occur very rapidly on newly emerged bean plants. The presence of a substrate suitable for growth (i.e., snap bean) of two strains of Pseudomonas syringae did not facilitate spread of these bacteria. Substrates that did not support growth (i.e., bare ground or soybean) of the P. syringae strains did not impede spread. The two P. syringae strains examined were known to differ in fitness on bean plants under field conditions. The strains appeared to be similar in ability to be dispersed. However, they differed in relative amounts of spread presumably due to their differential growth abilities on bean leaves.

Key words: Spread, dispersal, phyllosphere bacteria, Pseudomonas syringae.

INTRODUCTION

Any technological use of bacteria in an open system presents the potential risk that bacteria may move from the site where they are used to one where they may pose some risk of an adverse effect. Where bacteria that have the capacity to grow on aerial plant parts (the phyllosphere) may be used, for example for biocontrol, a number of factors may affect bacterial spread. In a typical agricultural or minimally managed ecosystem, a patchy array of different crops or other plants may be interspersed with areas where no plants exist.

Successful spread requires both dispersal to a new habitat and growth on that habitat. The literature suggests that rainsplash dispersal and aerosol dispersal are probably the most important means of spread of bacteria in the phyllosphere. If rainsplash and aerosol dispersal are the important modes of dispersal, then the likelihood of dispersal should diminish rapidly with distance. Ballistic particles generated by rainsplash travel only relatively short distances, and dilution of bacteria released from the phyllosphere in aerosols occurs very rapidly, and hence the likelihood of redeposition of aerosols decreases rapidly with distance from a source plant.

Given these assumptions, spread over a canopy of a plants on which rapid growth is likely should be facilitated. The bacteria may disperse any distance, grow on the leaves upon which they land, and redisperse to land on leaves somewhat farther from the original source. However, dispersal across the canopy of plants on which they only survive, but do not grow well, should allow redispersal of the original bacterial cells, but little amplification by growth. Finally, if the bacteria must cross bare ground, where most phyllosphere bacteria survive only poorly, and from which redispersal is less likely, only those dispersal mechanisms that move the bacteria the entire distance in a single hop should be expected to result in successful spread.

The fitness of a bacterium in the phyllosphere is reflected in the ability of that bacterium to grow in the phyllosphere. Thus, relative spread of two bacterial strains that differ in fitness but not in dispersal ability should provide insight into the relative contributions of dispersal and growth to spread.

Hence, the objectives of our research were i. to determine the impact of the nature of the substrate that occupies intervening space between a source (i.e., site of introduction) and sink (i.e., unintended site of introduction) on spread of phyllosphere bacteria, and ii. to determine the effect of bacterial genotype on the likelihood of spread.

EXPERIMENTAL DESIGN

Bacterial Genotype. The bacterial genotypes examined included two strains of Pseudomonas syringae pv. syringae (Pss) and a strain of a pink-pigmented facultative methylotroph (PPFM) of the genus Methylobacterium. Pss can grow quite successfully on leaves of snap bean (Phaseolus vulgaris) plants. Indeed, when population sizes of this pathogen become sufficiently large within the bean canopy, bacterial brown spot disease of snap bean occurs. Pss may survive quite successfully, but grow little on other plant species, such as soybean. The bacterium is not known to grow in soil, and survives poorly in soil and then only on plant debris. The two Pss strains were B728a, a highly virulent and field competent strain, and NPS3136, a non-lesion forming mutant of B728a. The inactivated gene in NPS3136 (lemA1::Tn5) is a regulatory gene (Hrabak and Willis, 1992). Although growth of NPS3136 on bean leaves in the growth chamber was found to be similar to the wild type parent, B728a (Willis et al., 1991), the mutant was less fit than its parent on bean plants under field conditions (Hirano et al., 1994). After two weeks, numbers of the mutant were approximately 50-fold lower than those of the parent, where each of the strains had been applied to separate field plots (Hirano et al., 1994). We expected, however, that the mutant should not be impaired in dispersal. The life strategy of the nonpathogenic PPFMs is very different from that of Pss.

Barriers. The intervening substrates (i.e., barriers) between snap bean sources and sinks were: i.snap bean (growth of Pss strains) ii. soybeans (growth of Pss strains limited or unlikely, but survival likely), and iii. bare ground, (neither growth nor survival and redispersal should be favored).

Plot Design. Each plot consisted of three concentric hollow squares (Figure 1). The outer square, the source, was 30 m on a side. The outer 6 m of this square was planted with inoculated snap bean seed to serve as the source for bacterial spread. The outer 6 m of the second square, 18 m on a side, served as the barrier, and the inner square, 6 m on a side, was planted to snap beans and served as the sink. There were three replicate plots for each bacterial genotype--barrier zone combination planted in a randomized complete block design with 5 m between plots. The sinks and barriers were planted on July 13, 1994. A thunderstorm that occurred on July 13 delayed planting of the sources until July 15 (B728a and NPS3136 sources) and July 16 (PPFM sources).

Sampling. Several precautionary measures were taken to eliminate the possibility of spread due to samplers: Leaves were always dry at the time of sampling, the sinks and barriers were entered only through specific paths (Figure 1), sinks were always sampled first, then barriers, and sources last. At each sampling time, 10 samples were taken from each source, 8 or 16 from each barrier, and 10 from each sink for processing by dilution plating. Roughly equivalent numbers of samples were also imprinted onto appropriate media for a rapid, qualitative assessment of spread.

Estimation of Bacterial Population Sizes. Each sample was individually homogenized with a polytron, serially diluted, then plated onto King s Medium B (KB) (King et al., 1954) supplemented with cycloheximide (100 µg/ml) and rifampicin (50 µg/ml) and nalidixic acid (50 µg/ml) for enumeration of B728a or rif and kanamycin (30 µg/ml) for NPS3136. A minimal medium supplemented with methanol and rif was used to enumerate PPFMs. Bacterial population sizes were log₁₀-transformed prior to analysis. The limit of detection of the dilution plating assay was approximately 2.27 log CFU/sample.

RESULTS AND DISCUSSION

Overall. Rapid spread of B728a and NPS3136 occurred during the short time (approximately 16 days) from emergence of plants in the sources to 21 days after planting (DAP). The frequencies of detection of both strains in their respective sinks are summarized in Figure 2. The PPFM strains applied to the seeds apparently did not colonize developing seedlings successfully. At all sampling times, the PPFMs were detected only at a very low frequency in the sources to which they had been applied. Further results with the PPFMs are not reported here.

First Detection of Spread. At 7 DAP, the first sampling time, the seedlings in the sources harbored large populations of B728a and NPS3136 (Figure 3). Unexpectedly, population sizes of NPS3136 were generally larger than those of B728a on the source plants. This may have been due to differential survival between application and planting. However, in other experiments, NPS3136 has survived and grown at least as well as B728a on developing seedlings up to the time of emergence. Both strains were detected at a low frequency on primary leaves collected from the sink areas (Figures 2, 3). The plants in the sources at 7 DAP were very small; many still in the process of emerging. Hence, spread was detected within two days or less after bean seedlings in the source areas had begun to emerge. The overall detection frequencies at 7 DAP for samples processed by imprinting as well as dilution plating (n = 450) were:

by strain, pooled across all sinks regardless of barrier type:

B728a = 2.7%, NPS3136 = 1.8%; and

by barrier type, pooled across all sinks regardless of strain:

snap bean = 0.44%, soybean = 1.55%, bare ground =

44%.

(None of the above differences were significant at the 5% level.)

The observation that both strains were detected in sinks surrounded by bare ground or soybean indicated that the absence of a substrate favorable for growth of the Pss strains was not impeding spread, at least at the scale examined (i.e., 6 m minimum distance, source to sink).

Neither B728a nor NPS3136 was detected on samples from the barriers. This was probably due to chance, as might be expected from the smaller number of samples collected from barriers compared to sinks and low frequencies of occurrence.

Increases in Amounts of Spread With Time. The frequencies of detection of B728a and NPS3136 pooled over all sinks were roughly 50% and 100% by 14 and 21 DAP, respectively (Figure 2). Bacterial population sizes on leaves from all areas are presented in Figures 4 and 5.

Effect of Barrier Zone. By 14 DAP, the nature of the barrier zone had a significant effect (p = 0.042) on spread of both strains (Figure 4). Interestingly, the largest rates of spread for both strains were to sinks surrounded by bare ground, not snap bean as had been anticipated. Had either or both of aerosol or rainsplash dispersal been important in spread of the bacteria from source to sink, the snap bean barrier should have facilitated spread. It did not. Indeed, the presence of a snap bean barrier, even a very young one, apparently slowed spread. Although this is difficult to rationalize if either aerosol or rainsplash dispersal were the principal mode of dispersal, it appears to be consistent with spread vectored by insects. If insects were moving the bacteria from source to sink, an insect that prefers snap beans to soybeans or bare ground might be expected to stop less frequently on soybeans or bare ground than on snap bean barriers. Thus, the effective distance that insects might disperse the bacteria in a single flight would be shorter, and the rate of spread slower over snap beans

By 21 DAP, there was no significant effect of the barrier zone on spread of B728a (Figure 5). Indeed, population sizes of B728a were essentially the same on sinks and snap bean barriers as in the sources. The ability of the bacterium to grow in the phyllosphere, not the nature of the barrier, was dominant in determining population sizes of B728a on leaves in the sinks. Additionally, the rapid colonization of nearly all of the leaves in the sinks from the relatively few leaves initially infested (presumably by insects) may reflect the effect of the more frequent shorter distance dispersal by rainsplash or aerosols.

Effect of Bacterial Genotype. Although the population sizes of NPS3136 were generally smaller than those of B728a in the sinks at 14 DAP (Figure 4), the differences were not statistically significant. However, by 21 DAP (Figure 5), there was a significant effect (p < 0.001) of bacterial genotype on spread. Population sizes of B728a and NPS3136 in the sinks (pooled over all barrier types) were 7.26 and 5.15 log CFU/leaflet, respectively. It is likely that the greater amount of spread observed with B728a compared to NPS3136 was due to a greater extent to differential abilities to grow and survive rather to be dispersed. Note that B728a and NPS3136 were detected at equivalent frequencies in the sinks at weeks 1, 2, and 3 (Figure 2), and hence appeared to be similar with regard to likelihood of dispersal. The decreased fitness of NPS3136 relative to B728a was also evident from lower population sizes in the sources at 14 (Figure 4) and 21 (Figure 5) DAP.

Conditions prevalent during the conduct of the experiment described here were highly conducive for growth of Pss. Rain was frequent and intense. The soil was kept wet by rain from planting to emergence--a condition that facilitates infestation of emerging primary leaves. Frequent rains during the first two weeks after plant emergence undoubtedly triggered rapid growth of the bacteria on the primary and trifoliolate leaves as they emerged. The very large population sizes thus achieved by B728a and NPS3136 on source plants provided abundant bacteria for spread to sink plants. Overall then, the experiment may provide an example of spread of Pss under 'ideal' conditions for growth and spread of this bacterium.

REFERENCES

Hirano SS, Ostertag EM, Savage SA, Willis DK, Upper CD (1994) Contribution of the regulatory gene lemA to fitness of Pseudomonas syringae pv. syringae in the phyllosphere and spermosphere under field conditions. Mol Ecol 3:607.

Hrabak EM, Willis DK (1992) The lemA gene required for pathogenicity of Pseudomonassyringae pv. syringae on bean is a member of a family of two-component regulators. J Bacteriol 174:3011-3020.

King EO, Ward MK, Raney DE (1954) Two simple media for the demonstration of pyocyanin and fluorescin. J Lab Clin Med 44:301-307.

Willis DK, Hrabak, EM, Rich, JR, Barta, TM, Lindow, SE, Panopoulos, NJ (1990) Isolation and characterization of a Pseudomonas syringae pv. syringae mutant deficient in lesion formation on bean. Mol Plant-Microbe Interact 3:149-156.

Figure 1. Diagram of plot design. Arrow in right plot represents specific path taken to enter a sink.

Figure 2. Frequencies with which B728a and NPS3136 were detected in the sinks (regardless of barrier type) as a function of sampling time. The frequencies were estimated from data obtained by dilution plating.

Figure 3. Population sizes of B728a and NPS3136 at 7 days after planting (i.e., within 2 days after emergence of plants in the source areas). Source samples were entire above ground parts of individual bean plants. Sink and barrier samples were individual primary leaves. Values plotted are the mean and SE of 10 samples from each of three replicate plots. The 'snap, soy, and bare' designations for the source and sink areas refer to the barrier type for those plots. The designations for the barrier areas refer to bacterial population sizes on snap and soybean leaves.

Figure 4. Population sizes of B728a and NPS3136 at 14 days after planting (i.e., within 9 days after emergence of plants in the source areas). All samples were individual primary leaves. See legend to Fig. 3 for additional explanations.

Figure 5. Population sizes of B728a and NPS3136 at 21 days after planting (i.e., within 16 days after emergence of plants in the source areas). All samples were individual leaflets from first trifoliolate leaves. See legend to Fig. 3 for additional explanations.