Persistence of Bacillus thuringiensis subsp. kurstaki in Urban Environments following Spraying

ABSTRACT Bacillus thuringiensis subsp. kurstaki is applied extensively in North America to control the gypsy moth, Lymantria dispar. Since B. thuringiensis subsp. kurstaki shares many physical and biological properties with Bacillus anthracis, it is a reasonable surrogate for biodefense studies. A key question in biodefense is how long a biothreat agent will persist in the environment. There is some information in the literature on the persistence of Bacillus anthracis in laboratories and historical testing areas and for Bacillus thuringiensis in agricultural settings, but there is no information on the persistence of Bacillus spp. in the type of environment that would be encountered in a city or on a military installation. Since it is not feasible to release B. anthracis in a developed area, the controlled release of B. thuringiensis subsp. kurstaki for pest control was used to gain insight into the potential persistence of Bacillus spp. in outdoor urban environments. Persistence was evaluated in two locations: Fairfax County, VA, and Seattle, WA. Environmental samples were collected from multiple matrices and evaluated for the presence of viable B. thuringiensis subsp. kurstaki at times ranging from less than 1 day to 4 years after spraying. Real-time PCR and culture were used for analysis. B. thuringiensis subsp. kurstaki was found to persist in urban environments for at least 4 years. It was most frequently detected in soils and less frequently detected in wipes, grass, foliage, and water. The collective results indicate that certain species of Bacillus may persist for years following their dispersal in urban environments.

Bacillus thuringiensis subsp. kurstaki is a common organic pesticide used to control defoliating pests including the gypsy moth, Lymantria dispar. The gypsy moth is a major forest pest that is especially predominant along the eastern seaboard and in the Midwestern United States. Over the last 20 years, thousands of acres have been treated with B. thuringiensis subsp. kurstaki to suppress or eradicate gypsy moth populations. The bacterium is applied to foliage as a water-based slurry. B. thuringiensis subsp. kurstaki is not typically harmful to mammals; but its toxin, when ingested, is lethal to the Lymantria dispar caterpillar (19,22,28).
In a recent review, Greenberg et al. (7) determined that B. thuringiensis provides the best overall fit as a nonpathogenic surrogate for Bacillus anthracis for spore fate and transport based on pathogenicity, phylogenetic relationship, morphology, and comparative survivability to biocides.
B. thuringiensis and B. anthracis are both Gram positive and aerobic, and they both form metabolically inactive endospores in response to environmental conditions. B. anthracis, the etiological agent of anthrax, is a bacterium of considerable concern in the biodefense community, but it is difficult to study its behavior in the environment, particularly for the wide-area releases often postulated in terrorist scenarios. Outdoor pesticide releases of B. thuringiensis subsp. kurstaki can therefore provide insight into the environmental fate and transport of B. anthracis following a deliberate release.
The literature on the persistence of Bacillus spp. in the environment is incomplete and contradictory; however, previous studies have demonstrated that viable Bacillus spp. spores may persist and remain dormant in laboratories or rural environments for years to decades. Early research on B. anthracis indicated that spores could survive indefinitely in a dry environment, such as in dust, or on laboratory swabs or blood spots on clothing (32). Manchee et al. detected B. anthracis in soil where the agent had been dispersed 40 years previously (14). Hendriksen and Hansen found that B. thuringiensis subsp. kurstaki was persistent in bulk soil in cabbage plots for 7 years (9).
The literature on Bacillus spp. persistence in water is predominantly focused on B. anthracis for biodefense. Although these studies may not be representative of general environmental conditions (33), they indicate that B. anthracis can remain viable in chlorinated or pond water for 2 years (3,12).
It is difficult to use data from the existing literature, which was largely collected in laboratories and undeveloped areas, to assess the implications for an urban area after a biological attack. However, understanding persistence in the urban environment will be critical to formulating effective response, restoration, and recovery plans. To that end, under the auspices of the joint Defense Threat Reduction Agency-Department of Homeland Security Interagency Biological Restoration Demonstration, soil, surface, water, and vegetation samples were collected from two urban areas (Seattle, WA, and Fairfax County, VA), up to 4 years after spraying, and analyzed for the presence of viable B. thuringiensis subsp. kurstaki.
Sampling of historic spray areas (1 to 4 years after spraying) occurred in the Seattle, WA, area. However, in 2008 (the year of this experiment), the Washington State Department of Agriculture did not identify a need to spray for gypsy moth. For that reason, Fairfax County, VA, was sampled immediately after spraying and then at intervals for up to 1 year. The results are reported below.

MATERIALS AND METHODS
B. thuringiensis subsp. kurstaki application. Fairfax County, VA, delivered one application of a commercial formulation of B. thuringiensis subsp. kurstaki (Foray 76B; Valent Biosciences, Libertyville, IL) via helicopter at a rate of 470 liters per km 2 . Since the manufacturer optimizes their product for toxin activity and does not measure the B. thuringiensis subsp. kurstaki concentration in the final formulation, a sample of the spray suspension was obtained from Fairfax County authorities for laboratory analysis. In Seattle, Foray 48B or Foray XG (Valent Biosciences, Libertyville, IL) were applied aerially or by ground as a 2% suspension on at least three successive occasions. According to the records, application rates ranged from 10 to 1,000 liters per km 2 . Since no archived spray material was retained, samples of Foray 48B and Foray XG were obtained from the manufacturer for laboratory analysis.
Sample locations and design. Sample collection occurred in 2008. One spray area in Fairfax County, VA, was sampled immediately before (background) and after B. thuringiensis subsp. kurstaki spraying and then at 6, 12, 24, and 48 weeks after spraying. Fairfax County's 2008 spray areas are shown in Fig. 1; the spray area sampled was designated block 35. Block 35 covers 0.737 km 2 and is largely residential. During each sampling interval, including background, a total of 245 samples were collected from this spray block. In Seattle, WA, samples were collected in locations that the Washington State Department of Agriculture sprayed between 2007 and 2004, Fig. 2. The Seattle spray blocks ranged from 0.02 to 0.4 km 2 and were predominantly commercial or mixed commercial/residential areas with tree canopy cover Ͻ15%. Detailed information on Seattle spray block and B. thuringiensis subsp. kurstaki application parameters is available from the Washington State Department of Agriculture (http://agr.wa.gov/PlantsInsects/InsectPests/GypsyMoth /ControlEfforts/PastControlEfforts/PastControlEfforts.aspx). A control area that had not been sprayed by the Washington State Department of Agriculture was also sampled to determine the background presence of B. thuringiensis subsp. kurstaki in the Seattle area (Fig. 2). Tables 1 and 2 show the spray block size and the timeline for sample collection in each location. Between 218 and 240 samples were collected from each spray block, with a total of 242 samples collected from the control block.
Three sampling schemes were used to characterize each spray area: probabilistic, close, and targeted (15,30). Initial sample numbers were chosen to give 99% confidence that at least 95% of the area was without detectable spores provided all samples were negative. A grid or transect technique was used to define the probabilistic sample locations; a 9.1-by-9.1-m grid was used to define sample locations in relatively homogeneous areas, such as parking lots, while a smaller, 2.7-by-2.7-m grid was used to define sample locations in heterogeneous areas such as residential neighborhoods. Probabilistic sampling implies uniform distribution of the agent interrogated; to test this assumption, a set of "close" (or secondary) samples were collected for a randomly chosen 10% of the probabilistic samples at each site. Close samples were of the same type as their associ-  ated sample and were collected one foot to the north of the original sample whenever possible (the results of the "close" samples are not analyzed explicitly in this article; as predicted, the assumption of uniform distribution was incorrect). Finally, a set of targeted samples were collected in locations where B. thuringiensis subsp. kurstaki was likely to persist based on literature information (e.g., shady or damp locations; standing water). An additional 10% of the total number of samples was collected as field blanks at each site. Sample collection. Sample collection was performed using the protocols contained in the BioWatch Outdoor Program Guidance (29). Samples from Fairfax County, VA, included soil, wipe, water, grass, and leaf samples. Samples from Seattle, WA, included soil, wipe, and water samples. Vegetation was not collected in Seattle; due to the longer duration between spraying and sample collection, it was assumed the bacteria would have been washed off the foliage.
In brief, the collection of soil samples involved scraping any organic material off the soil surface with a sterile spatula and collecting enough solid soil from the top 5 cm to fill a 50-ml conical tube up to the 30-to 50-ml level. Using a sterile pipette, 10 ml of standing outdoor water (e.g., puddle or lake) was collected into a 50-ml conical tube. Grass was cut using sterile scissors, with enough material collected to fill a 50-ml conical tube up to the 30-to 50-ml level, and leaves were cut from trees using sterile scissors with enough material collected to fill a 50-ml conical tube up to the 30-to 50-ml level (leaves that had fallen on the ground were not collected). Wipe samples were collected using sterile rayon gauze (7.6 by 7.6 cm; Dynarex Corp.) premoistened with 3 ml of phosphate-buffered saline (Fisher Scientific, Pittsburgh, PA). An area of ϳ1 m 2 was wiped using vertical then horizontal S strokes; the wipe was then placed in a 50-ml conical tube. All conical tubes were capped and sealed with Parafilm and placed in a self-sealing bag. The bag was wiped with a hospital grade bleach wipe (Dispatch; Caltech Industries, Midland, MI), allowed to air dry, and then placed on ice in a cooler for transport to the analytical laboratory.
Sample preparation and DNA extraction. For soil samples, 0.4 g of the sample was used for DNA extraction. For vegetation samples, a quarter of the sample was washed with 10 ml of PET buffer (100 mM sodium phosphate buffer with 10 mM EDTA and 0.01% Tween 20; Fisher Scientific), and an aliquot of the wash was used for DNA extraction. For wipes, the entire sample was washed with 2 to 3 ml of PET buffer, and an aliquot of the wash was used for DNA extraction.
DNA was extracted from soil and aliquots of vegetation and wipe washes using the commercial FastDNA spin kit for soil (MP Biomedicals LLC, Solon, OH). Extracts of the same type (e.g., soil) were pooled together by location (i.e., samples collected in close proximity) to minimize analysis time and cost. Each pool combined a maximum of three extracts. Extracts were then subjected to a molecular screen using B. thuringiensis subsp. kurstaki-specific real-time PCR.
Screening analysis (molecular screen). PCR inhibitors such as humic acid and cellulose are common in environmental samples. To test for inhibition following extraction, each batch of samples was tested for the presence of inhibitors by performing real-time PCR amplification of 16S ribosomal DNA (rDNA). Regardless of the source, environmental matrices such as soils typically contain large amounts of 16S rDNA. If no amplification was observed from the soils, which were expected to contain large amounts of target material, it was determined that inhibitors were likely present. When amplification of the 16S rDNA was unsuccessful, the batch of samples was diluted to 1:10 or 1:100 with DNAand DNase-free water and tested again to ensure amplification prior to running B. thuringiensis subsp. kurstaki-specific real-time PCR.
DNA extract pools were tested for the presence of B. thuringiensis subsp. kurstaki DNA using real-time PCR with two B. thuringiensis subsp. kurstakispecific primers designed as LANL. The sensitivity, specificity, and the ability of these primers to detect B. thuringiensis subsp. kurstaki DNA in a wide matrix of samples were tested (see the supplemental material). Both 16S and B. thuringiensis subsp. kurstaki real-time PCR assays used the Bio-Rad iQ SYBR green Supermix (Bio-Rad, Hercules, CA) and LANL-designed 16S and B. thuringiensis subsp. kurstaki PCR primers (Invitrogen, Carlsbad, CA) in 25-l reactions that contained 50 mM KCl, 20 mM Tris-HCI (pH 8.4), 0.2 mM concentrations of each deoxynucleoside triphosphate (dATP, dCTP, dGTP, and dTTP), iTaq DNA polymerase at 25 U/ml, 3 mM MgCl 2 , SYBR green I, 10 nM fluorescein, stabilizers, 7.5 nM forward and reverse primers (B. thuringiensis subsp. kurstaki or 16S), and 5 l of DNA extract (diluted or undiluted). The PCR cycling conditions were as follows: a melting temperature of 95°C for 2 min, followed by 45 cycles of melting at 95°C for 15 s, annealing at 60°C for 1 min, and extension at 70°C for 10 s. Real-time PCR was conducted in a Bio-Rad iCycler iQ real-time PCR detection system.
Results were reported as threshold cycle (C T ) values. The C T values were converted to an estimate of target genome copy number based on a standard curve generated for each plate of samples using serial dilutions of commercially available B. thuringiensis subsp. kurstaki DNA (American Type Culture Collection [ATCC], Manassas, VA). The threshold for selection of a sample for viability analysis was Ͼ1,000 target genome copies for either primer. Each plate subjected to real-time PCR contained the following controls: (i) a no-template control to ensure there was no background contamination; (ii) a panel of controls with known concentrations of B. thuringiensis subsp. kurstaki DNA (obtained from ATCC, Manassas, VA) in the form of genome copy number (10 0 to 10 6 ) to enable conversion of C T values to estimated genome copy number for each reaction; and (iii) a set of reactions comprising a diversity panel that included DNA from B. thuringiensis subsp. kurstaki near neighbors and unrelated organisms to ensure the specificity of the assay. Viability analysis. Pooled or individual samples that had an estimated B. thuringiensis subsp. kurstaki genome copy number of Ͼ1,000 target copies for either of the two primer pairs were processed for viability. In addition, at least 10% of samples that were negative by the molecular screen were assessed by plate culture to determine the reliability of the screen. To do this, fresh aliquots of the corresponding unpooled soil, water, vegetation, or wipe samples were washed with Trypticase soy broth (TSB; Becton Dickinson/Fisher Scientific). The supernatant was then removed (1.5 ml for soil samples and 200 l for water, vegetation, and wipe samples) and repooled using the same pooling scheme used for PCR. Pooled samples were heat treated (80°C for 10 min) to kill all vegetative cells. Then, 100 l of the pooled sample was added to Trypticase soy broth (broth culture), or three serial dilutions (10 0 to 10 Ϫ2 ) of each pooled sample were plated on Trypticase soy agar medium (Becton Dickinson/Fisher Scientific) containing cycloheximide (50 mg/liter; Sigma-Aldrich, St. Louis, MO) to inhibit fungal growth. Incubation occurred at 36°C for 16 h for broth culture and 24 h for plate culture (11). Broth cultures were observed for turbidity, and plates were observed for colonies with B. thuringiensis morphology. A culture result of growth or no growth was reported.
To confirm that broth cultures contained viable B. thuringiensis subsp. kurstaki, 50-l aliquots of the broth were taken at t ϭ 0, before incubation and then at t ϭ 16 h. The aliquots were subjected to heat lysis to make DNA available for B. thuringiensis subsp. kurstaki-specific real-time PCR using each of the two primers used in the molecular screen. Viable B. thuringiensis subsp. kurstaki was confirmed to be present if the ratio of the estimated target genome copy number at 16 h compared to zero hours was greater than 10.
To confirm that colonies on the plates were from viable B. thuringiensis subsp. kurstaki, a loopful of material was scraped from the 10 0 dilution plate of each positive sample and suspended in 50 l of distilled water. This material was subjected to heat lysis and subjected to B. thuringiensis subsp. kurstaki-specific real-time PCR using the same two primers used in the molecular screen. The results were reported in the form of C T values. Analysis of spray suspensions of B. thuringiensis subsp. kurstaki (Foray76B, 48B, and XG) was also conducted by plate culture as described above. An aliquot of the spray suspension was diluted 1,000-fold in phosphate-buffered saline (PBS) and heat treated. Serial dilutions of this preparation (100-l aliquots) in PBS were plated, and colony counts were obtained after overnight incubation. Triplicate measurements were made, yielding average values.
Quality assurance and control. Quality assurance was implemented for both sample collection and analysis to ensure high confidence in the data.
Sample collection. All samples were collected by personnel wearing clean, disposable nitrile gloves and booties. Gloves were changed after collection of each sample. Each sample was individually bar-coded and chain of custody was electronically tracked from collection through analysis using custom field software and a laboratory information management system (LIMS). Field blanks, analyzed to ensure no contamination occurred during sample collection, comprised 10% of the samples collected.
Sample analysis. During laboratory analysis, samples and extracts were contained and manipulated in closed conical, centrifuge, and microcentrifuge tubes and/or sealed deep well blocks to minimize aerosolization and cross-contamination. A positive control consisting of a soil sample spiked with a known concentration of B. thuringiensis subsp. kurstaki spores was included in each batch of samples processed to ensure performance of the assays. A total of 10% of all samples were processed as duplicates in the laboratory to assess internal consistency. All samples that had an estimated B. thuringiensis subsp. kurstaki genome copy number of Ͼ1,000 copies were processed for viability and, in order to test the effectiveness of the PCR screen, Ͼ10% of all samples that failed the PCR screen were also processed for viability.
The quality of DNA extracted is dependent on the matrix and the extraction process used. The DNA extraction kit used in this project was specific for DNA extraction from soils. Other matrices sampled in the present study presented less of a challenge for extraction. To ensure that the DNA extraction process was successful, soil extracts were tested for the presence of DNA by PicoGreen with the Quant-iT PicoGreen dsDNA assay kit from Molecular Probes (Invitrogen, Carlsbad, CA). Soil extracts were used since they were expected to contain the most DNA and would therefore show high levels of fluorescence when stained with Picogreen. Two reference soil samples that represented different soil types, sand and clay, were spiked with a known concentration of B. thuringiensis subsp. kurstaki spores and processed with each batch. If the reference extracts showed negative results, then the DNA extraction process was repeated for that batch of samples.

Spray suspension analysis.
Samples of each commercial B. thuringiensis subsp. kurstaki spray formulation were obtained, and viable spore concentrations were determined. A concentration of 1.3 ϫ 10 9 Ϯ 5 ϫ 10 8 CFU of B. thuringiensis subsp. kurstaki per ml of suspension was measured from samples provided by the Virginia Department of Agriculture. A concentration of 8.3 ϫ 10 8 Ϯ 1 ϫ 10 7 was measured from samples of Foray 48B and a concentration of 6.0 ϫ 10 9 Ϯ 3 ϫ 10 8 was measured from samples of Foray XG (provided by Valent Biosciences, since the Washington State Department of Agriculture does not maintain samples of historical spray material).
Based on the results of culture analysis and the application rate provided by Fairfax County, ϳ5 kg of B. thuringiensis subsp. kurstaki was applied to spray block 35 (Table 1). Table  1 also includes an estimate of the amount of B. thuringiensis subsp. kurstaki remaining at the time of sampling, based on literature half-life values of 100 and 200 days (19).
Using the results of analysis of B. thuringiensis subsp. kurstaki spray material representative of the spray formulation the application rate provided by Washington State, the estimated amount of B. thuringiensis subsp. kurstaki applied in each spray block sampled is given in Table 2. Approximately 2 kg of B. thuringiensis subsp. kurstaki was applied to Bellevue and Eastlake, approximately 10 kg of B. thuringiensis subsp. kurstaki was applied to Madison, and less than 100 g was applied to Rosemont and Kent. Table 2 also includes an estimate of the amount of B. thuringiensis subsp. kurstaki remaining at the time of sampling, based on literature half-life values (19).
Environmental sample analysis. The numbers of sample pools and field blanks analyzed for each spray area are presented in Table 3, with the time elapsed between B. thuringiensis subsp. kurstaki application and sample collection.
Urban soils were found to be the most reliable reservoir for viable B. thuringiensis subsp. kurstaki. Soil results from Fairfax County are presented graphically in Fig. 3; soil results from Seattle, WA, are presented in Fig. 4. The largest percentages of soil samples containing detectable DNA and viable cultures were obtained immediately after spraying in Fairfax County (t ϭ 0 in Fig. 3  Other sample types were less reliable reservoirs of B. thuringiensis subsp. kurstaki compared to soil and exhibited more variability. Table 4 presents wipe, water, grass, and leaf results. Wipes were collected from all locations and from a number of urban surfaces, including, but not limited to, concrete, asphalt, and metal (e.g., manhole covers). No wipe, water, grass, or leaf samples collected in the Fairfax County background samples or the Seattle control area were found to be positive by PCR or culture.
Viable B. thuringiensis subsp. kurstaki was obtained from wipe samples in Fairfax County at all time points except at t ϭ 0. The highest percentage of culturable wipes was collected at in Fairfax County at t ϭ 48 weeks, although no wipes from that sample set passed the PCR screen. Similar variability was ob- Grass and leaf samples were only collected in Fairfax County, and the results were similarly inconsistent. Few grass and leaf samples from Fairfax County obtained after t ϭ 0 passed the PCR screen. However, viable B. thuringiensis subsp. kurstaki was detected in grass and leaves at t ϭ 0, 6, and 12 weeks (Table 4).

DISCUSSION
The purpose of this study was to determine how long B. thuringiensis subsp. kurstaki persists in urban environments at detectable levels. B. thuringiensis subsp. kurstaki was found to persist for at least 4 years; data for the past 4 years were not collected. Sampling of historic spray areas (1 to 4 years after spraying) occurred in relatively small, well-separated locations in the Seattle, WA, area. The year sampling occurred in Seattle, the Washington State Department of Agriculture did not perform any new B. thuringiensis subsp. kurstaki spraying. For this reason, Fairfax County, VA, was sampled immediately after spraying and then at various intervals for up to 48 weeks.
Using the reported half-life of B. thuringiensis subsp. kurstaki in cabbage patches (100 to 200 days) (19), estimates of the amount of B. thuringiensis subsp. kurstaki remaining in the spray areas at the time of sample collection were made (Tables  1 and 2). There are many assumptions in these estimates, the most implausible of which is that all B. thuringiensis subsp. kurstaki was applied to and remained inside the spray area boundary (1,28). However, the estimates likely give a conservative upper bound of the amount of B. thuringiensis subsp. kurstaki remaining at the time of sampling. The results of this experiment are consistent with the overall trend indicated in Tables 1 and 2: positive samples were collected in decreasing numbers as the time after spraying increased, but positives were still observed 4 years after spraying.
The major difference between the estimates in Tables 1 and 2 and the results lies in the different trends observed in Fairfax County and Seattle. Fairfax County shows a more marked and less regular decline in the percentage of samples containing viable material over 1 year. Seattle shows a steadier but less steep decline. This may be due to several factors. First, significant flooding occurred within 1 week after spraying in Fairfax County. This may have transported part of the initial spray material out of the sampled area. Second, it may be due to climatic or environmental differences between the urban areas. Many factors have been postulated to contribute to Bacillus spp. persistence or decay, including regional climatic conditions, soil alkalinity, the presence of shale, sandstone, or limestone in the soil, and potential interactions with the rhizosphere or earthworms (5,9,24,32). These factors were not characterized in this experiment. Finally, it is also possible these and other anomalies in the data were an artifact produced by inherent uncertainties in the sample collection process. Environmental sample collection techniques are difficult to standardize between sample collection personnel, and individual variability may have contributed to the overall results. Because the present study investigated viability trends only and the results are qualitative (presence/absence) rather than quantitative, it is difficult to draw a conclusion on the significance of the decay trends.
Soil was the best matrix for recovery of viable B. thuringiensis subsp. kurstaki. It is difficult to determine whether this was because B. thuringiensis subsp. kurstaki was predominantly present in soils, compared to other matrices, because B. thuringiensis subsp. kurstaki was most efficiently collected in soils (e.g., soil volumes sampled were higher than wipe volumes), or because B. thuringiensis subsp. kurstaki was most efficiently extracted from soil samples or whether this was a combination of multiple factors. The results, however, are not inconsistent with previous studies on the persistence of Bacillus spp. in rural soils. Manchee et al. found that 13% of soil samples collected 40 years after release at a largely rural biological warfare testing plot contained viable B. anthracis (14). In our study, 15% of soil samples collected at an urban, highly trafficked Seattle location contained viable B. thuringiensis subsp. kurstaki after 4 years (Fig. 4). Several other studies have detected Bacillus spp. in soil following its deliberate application (12)(13)(14)23); the results indicate persistence up to 6 months in a Russian field (13) and up to 7 years in a cabbage plot in Denmark (26). Differences among these results may be explained by more or less favorable climatic and environmental conditions (5); however, the data in aggregate clearly indicate that soil is a favorable reservoir for viable spores, whether the soil is in a rural or urban environment.
Other sample matrices were less reliable reservoirs of viable B. thuringiensis subsp. kurstaki. Wipes produced more consistent results than water and vegetation, but no more than 26% of wipes in Fairfax County and 16% in Seattle were viable at any time point (Table 4). Although literature on the persistence of Bacillus on surfaces and in water is sparse, it does indicate a strong dependence upon surface type and climatic and environmental conditions (e.g., sunlight and moisture) (10,16). B. thuringiensis has been found to persist on vegetation for days in direct sunlight (10) to a year on spruce needles (22). One lab study recovered viable B. anthracis from canvas after 41 years (25), and another evaluated persistence on wood, laminate, aluminum, polyvinyl chloride, and other surfaces, and recovered viable B. anthracis for up to 987 days (4). Comparisons of these data to laboratory studies are not wholly appropriate since the transient nature of water and vegetation likely contributed to the inability to recover viable B. thuringiensis subsp. kurstaki from these matrices over longer periods of time. However, the results of the present study are consistent with literature reports from other experiments (10,22).
Secondary to demonstrating persistence in urban environments, the present study illustrates the differences between PCR screening and culture for environmental sample analysis. In some cases, culture analysis was more sensitive than the PCR screen, as demonstrated in samples from Fairfax County at t ϭ 12 and t ϭ 48 weeks (Fig. 3) and in Seattle from Eastlake 2005 and Bellevue 2004 (Fig. 4). In all cases, these samples contained small amounts of material (Ͻ1,000 genome copies by PCR), and the corresponding culture results were likely due to significant differences in the processing of samples for culture versus PCR: (i) larger aliquots of sample were used for culture than for PCR; (ii) DNA extraction efficiency for PCR is not 100%; and (iii) residual PCR inhibitors can affect the quantity of genome copies estimated.
In this experiment, both assays provided independent information on the presence of B. thuringiensis subsp. kurstaki, although viable cultures were also confirmed by PCR. While knowledge of viable agent by laboratory culture is important for assessing public health impacts in biological restoration efforts, it may not always be the best indication of presence of an agent in the environment. Bacillus spp. may be sensitive to sample collection methods, specific media, and laboratory conditions, all of which can render it viable but nonculturable in a laboratory setting (2,17). PCR allows rapid detection with specificity and sensitivity (18). On the other hand, poor extraction efficiency and the presence of inhibitors (e.g., humic matter) can reduce PCR's effectiveness (27,31). In this experiment, sample aliquots for viability testing were plated along with the sample matrix, allowing higher sensitivity of the viability assay in some cases; the results of soil samples from t ϭ 12 weeks at Fairfax County are an example (Fig. 3). This outcome was especially pronounced when PCR concentrations were near the predetermined cutoff threshold of 1,000 genome copies. It should also be noted that PCR detects free DNA and nonviable agent. When time and cost are not a constraint, ideally, both techniques should be used for analysis of environmental samples.
The goal of the present study was to provide information on the persistence of a near neighbor of B. anthracis to inform efforts to determine how to restore an urban environment following a biological attack. Analysis of B. thuringiensis subsp. kurstaki persistence from two urban areas, one on the east coast and one on the west, demonstrated B. thuringiensis subsp. kurstaki can be expected to persist in urban environments for prolonged periods of time. At 48 weeks after B. thuringiensis subsp. kurstaki was applied, 85% of soils, 26% of wipe samples, and 50% of water samples contained viable B. thuringiensis subsp. kurstaki in Fairfax County, VA. In Seattle, WA, 77, 53, 25, 23, and 15% of soil samples contained viable agent at 1, 2, 3, and 4 years, respectively, after the B. thuringiensis subsp. kurstaki spray events. Consistent with the literature, soil was a reliable reservoir for Bacillus spp. at both locations. The results obtained from other environmental matrices (surface wipes, water, and vegetation) were less easily interpreted, but viable samples at various time points provided additional evidence of persistence in these different matrices in urban areas. Since B. thuringiensis subsp. kurstaki is a reasonable surrogate for fate and transport studies (7), it can be inferred that B. anthracis may persist for several years if released in an urban environment.