Microbiological analyses of fruits and vegetables produced by farms in Minnesota and Wisconsin were conducted to determine coliform and Escherichia coli counts, and the presence of E. coli, Salmonella and E. coli O157:H7. During the 2003 and 2004 harvest seasons, 14 organic (certified by accredited organic agencies), 30 semi-organic (used organic practices but not certified) and 19 conventional farms were sampled to collect and analyze 2,029 pre-harvest produce samples (473 organic, 911 semi-organic, 645 conventional). E. coli prevalence was linked to certain farm management practices, collected through farmers’ surveys at the beginning of each of the two harvest seasons. Effect of the harvest months on E. coli prevalence was determined using generalized estimated equation model of logistic regression. The E. coli isolates were fingerprinted using pulsed filed gel electrophoresis, and the clonal diversity among isolates obtained from the same farm was determined. None of the produce samples had Salmonella or E. coli O157:H7 contamination in the two years of this study. E. coli contamination was detected in 8% of the samples, and leafy greens, lettuces and cabbages had significantly greater E. coli prevalence compared to all the other produce types in both years for the three farm types. Fertilization of produce with animal wastes increased the risk of fecal contamination significantly in both organic and semi-organic produce. Organic farmers who aged their animal manure for less than 6 months had 4-times greater risk of fecal contamination in their produce, compared those that aged for more than 6 months. The risk of contamination from the fecal indicator bacterium was significantly greater in June and July compared to August and September, irrespective of year of sampling, produce types and farm types. A wide diversity was observed among the E. coli isolates from the produce collected from different farms, and also among the isolates obtained from the same farm over the three years. These results have been presented at different scientific meetings and the most relevant results of the investigation have been sent to the participating farmers.
Fresh fruits and vegetables, including fruit juices are essential components of the regular diet of any human group. Numerous evidence of health and nutritional benefits from the consumption of fresh fruits and vegetables has been documented in the literature (9, 14, 21). According to most agricultural economists, the sales of fresh-cut produce have increased sharply from approximately $3 billion in 1994 to $12.5 billion in 2004 (42). The U. S. Department of Agriculture (USDA) has recently emphasized the need for consumption of fresh produce by recommending at least five daily servings in the diet (6).
In recent years, the sales of organic foods have increased at an annual average rate of 20% in the U. S., and most estimates suggest that the market expansion for organic foods will continue at the same rate for the next 5 years (8, 33). As much as 42% of the organic food sales are contributed by organic produce and 93% of which is in the form fresh fruits and vegetables (8). The USDA’s Organic Rule implemented in 2002, included the acceptable production practices for foods marketed as “organic” which largely limited the use of fertilizers to animal and plant wastes for vegetable crop production (32).
In recent years, the number of foodborne outbreaks caused by contaminated fresh fruits and vegetables increased sharply. Produce accounts for 12% of foodborne illnesses and 6% of foodborne outbreaks today in the U. S., compared to 1% and 0.6%, respectively, in the 1970s (5). A report by Sivapalasingam et al. (36) suggested that the majority of produce-related foodborne outbreaks, for which the etiological agent(s) were identified, were caused by pathogenic bacteria. In the last 10 to15 years, Salmonella and Escherichia coli O157:H7 have been the two most common etiological agents responsible for produce-related outbreaks in this country. Published reports have documented outbreaks of Salmonella and E. coli O157:H7 infection from produce consumption in various states in the U. S., including Minnesota (13, 26).
Because organic growers rely primarily on animal manure for fertilization of their soil, it has been suggested that organically grown foods have a greater risk of pathogenic contamination, compared to their conventional counterparts (38). However, there are very few published reports that have conducted microbial risk assessment studies of organic fruits and vegetables. In our previous study, the prevalence of Escherichia coli in certified organic produce at the pre-harvest stage was greater than their conventional counterpart, but this difference was not statistically significant (29). That study also reported that none of the pre-harvest certified organic and conventional produce tested positive for either Salmonella or E. coli O157:H7 and only two semi-organic samples had Salmonella contamination. Another study found similar levels of E. coli contamination in organic and conventional spring mix, and all samples were negative for Salmonella and Listeria monocytogenes (34). The issue whether organic produce poses a greater risk for foodborne disease, however, remains largely unresolved.
Numerous reports have documented that animals such as cattle, sheep, pig and chicken are major reservoirs of foodborne pathogens such as E. coli O157:H7 and Salmonella (12, 24, 40). When farmers use manure from these animals for fertilization of produce plants, these pathogens might get transmitted to fruits and vegetables harvested from those plants (3). Such transmission was reported in lettuce when manure, inoculated with laboratory cultures of E. coli O157:H7, was used for fertilization of (37). Composting is an exothermic microbiological process, which generates high temperatures like 55 to 65C under proper conditions of aeration, moisture, particle size and carbon-to-nitrogen ratio for long enough duration that inactivates these foodborne pathogens (10). Published reports have documented the effectiveness of composting in destroying E. coli O157:H7 and Salmonella in cow manure (23). The USDA has specified composting techniques that organic growers are required to follow to minimize the risk of contamination from manure-based fertilizers (32). The USDA has also specified fertilization-to-harvest intervals that would reduce the chances of pathogen survival in non-composted manure. A few reports evaluated the effectiveness of such duration in minimizing risk of contamination in vegetables (16).
A few surveys focusing on microbial quality of organic produce have also been reported in the literature. All these surveys focused on retail and post-harvest samples of organic vegetables, and none of these studies reported the presence of pathogenic bacteria such as Salmonella and E. coli O157:H7 in organic produce (22, 25, 35). Outbreaks of infection from foodborne pathogens in contaminated organic produce have not been documented in the U. S. to date. The generation of data on comparative microbiological quality and safety of organic and conventional produce at farms would greatly enhance our understanding of factors that contribute to contamination of fresh fruits and vegetables. The reports available in the literature have documented the survival of contaminating bacteria in manure and in manure-amended soil (7, 15). Other laboratory and garden-scale studies have evaluated the effects of manure treatment and application to soil on the survival of pathogenic and fecal indicator bacteria in agricultural soil and vegetables grown in such soils (17).
In the United States, the number of foodborne outbreaks from produce consumption reaches the maximum during the summer months, while in the winter, such outbreaks occur less frequently (31). In Minnesota, a consistent seasonality pattern in the number of outbreaks of foodborne Escherichia coli O157:H7 infection has been reported (2). Published reports on microbiological quality of fresh fruits and vegetables have not focused on variation in contamination during different months of the year. In a study on organic leaf lettuce in Norway, the samples were collected in July, August, September and October, 2001 (22). They reported counts of thermo tolerant coliform and E. coli, and E. coli prevalence in the samples collected during those months, but it was a study with one year of sampling, and did not report variation in the counts and prevalence across the four months of the study. In another report, microbiological quality of organic and conventional spring mixes were compared (34). Sampling was done in one year, during the four months of April to August. Like the previous report, that study also did not focus on variation in microbiological counts during the four months. A larger study involved 398 samples of leafy greens, herbs and cantaloupes collected from farm during a longer period of November 2000 to May 2002 (20). Although this report involved samples collected during a period more than a year, they did not report the variation in bacteria; counts and prevalence during these months of the study.
Recently, Ishii et al. (18) and Byappanahalli et al. (4) reported the survival and subsequent naturalization of Escherichia coli in temperate soils of watersheds of various lakes. Both of these reports determined that E. coli can persist in these environmental niches for months, and even for a year, and eventually become a natural habitat of the environment. The persistence of E. coli in an environment such as farms producing fruits and vegetables has not been studied.
A. Project Objectives
The specific objectives of this proposal are to:
1) Determine the presence of fecal indicator organisms (coliforms, Escherichia coli) and pathogens (E. coli O157:H7, Salmonella) in organic and conventional fruits and vegetables produced by farmers in Minnesota and Wisconsin at the preharvest stage.
2) Conduct trace-back investigations in participating organic farms by comparisons of bacterial strains isolated from environmental samples and those isolated from produce
3) Identify potentially high-risk management practices and provide recommendations for improvement
4) Disseminate results and findings among the agricultural community.
B. Project Outcomes
1) A quantitative and comparative microbial risk assessment of fresh fruits and vegetables produced by organic farms.
2) A series of improvements in management practices such as modifications in manure type, proper use of composting and time of application of organic fertilizers that will eventually reduce the risk of microbial contamination and produce safer fruits and vegetables.
3) Enhanced farmers’ awareness regarding application of manure fertilizer, irrigation water and the rationale for composting requirements.
4) Increased confidence of those farmers already using practices linked to low microbial loads and better safety of their produce.
5) Strengthen organic agriculture by providing the basis for enhanced consumer confidence that might lead to increased demand.
6) Help solving the long-standing debate on the microbial safety of the organic fruits and vegetables for the scientific community and the society in general.
(a) Classification of farms
Farmers that grew fresh fruits and vegetables were invited to participate in this study by telephone or by personal contacts. All participating farms were located in Minnesota and Wisconsin. The farmers participated by allowing the collection of fruits and vegetables from their cultivation plots before harvest. Farms were classified in three major categories: Organic, Semi-organic and Conventional. Organic farms were those that were currently certified by an USDA-accredited organic certification agency. Semi-organic farms were those that reported using organic practices but were not certified. Conventional farms were those operations that could use any type of farm inputs and practices. Participation from organic farmers was less in the 2nd year of the study compared to the 1st. The main reason that we found from communications with organic farmers in Minnesota and Wisconsin was that some of the organic growers did not renew their certification. Most of the organic farmers in the upper Midwestern states of Minnesota and Wisconsin are small-scale, family farmers, and some of them were not interested to conform to the necessary paperwork and fees associated with the renewal process.
A questionnaire was mailed to all the participating semi-organic, organic and conventional farmers at the beginning of each harvest seasons. General information such as names, contact information, years in business, market, certification status and produce types grown was collected. The major focus of the questionnaire, however, was use of animal waste and other pre-harvest production and handing practices used by produce growers. The relevant study questions are shown in Table 3.1. A stamped return-envelope, in which the farmer’s responses were received, was attached with the questionnaire.
(b) Sampling of fresh fruits and vegetables
The participating farms were visited from 2 to 3 times during June, July, August and September, in the year of 2003 and 2004. Samples of fruits and vegetables were collected directly from the farm fields, and without any washing or rubbing-off of soil particles, the samples were transferred into sterile zip-lock bags. The gloves and knives were sanitized with alcohol swabs between collections in order to prevent cross-contamination. Produce types included the following: lettuces, leafy greens, cabbages, peppers, tomatoes, berries, broccoli, summer squash, cucumber, zucchini, and other types of produce in small numbers e.g. bok choi, cantaloupe, apple, kohlrabi, sprouts and peas. Types of lettuce included such as Romaine, head lettuce and leaf lettuce. Leafy greens were spinach, kale, collard, Swiss chard and mixed greens. Chinese and red cabbages were the two types of cabbages collected. Pepper types included bell, banana, yellow and red peppers. Types of tomatoes included Roma, cherry and beefsteak tomatoes. Berries were strawberry, raspberry and blueberry.
The sample amounts varied from one produce type to another. For leafy greens like spinach, kale, collard, about 500 g of leaves were collected from different plants in a portion of a field of cultivation, and those leaves constituted one sample. For produce types such as cucumber, summer squash, pepper, tomato, from 2 to 5 of these vegetables were collected from different parts in a particular location of a field to constitute a sample. For produce types such as cantaloupe, cabbage, bok-choi, and head lettuce, one head of lettuce or cabbage or bok-choi or one cantaloupe constituted one sample. For fruits like raspberry and strawberry, about 500 g of fruits, ripe enough for harvest, from different plants in a particular location of the field were collected in sterile plastic boxes. For representative sampling from a field of cultivation, produce from different locations of a field were sampled. Most of the produce fields were sampled in triplicate, and for large fields of cultivation, as many as five samples were collected from different locations of the fields.
Sample bags and boxes were properly marked by recording produce type, farm identity, sample number, and date of collection on them. The sample bags were then placed in insulated coolers with ice-packs and were sent to the laboratory. Samples were received within 10 hours of collection, and were stored at 4C until microbiological analyses began. Samples were stored in insulated coolers, and in some cases in card-board boxes. Analyses of produce samples started within 48 hours of sampling.
(c) Microbiological analyses
Microbiological analyses of fruits and vegetables started with sample preparation, which depended on the type of fruits and vegetables. For leafy greens such as lettuce, spinach, kale, collards, cabbage and bok-choi, leaves from the outside and inside sections of the produce were used to make up 25 g of sample. For produce types such as summer squash, zucchini, cucumber, tomato, pepper, one or two of these vegetables were cut into small pieces, and pieces from different portions of the vegetable were used to make up 25 g of sample. For fruits such as raspberry and strawberry, 2 to 6 (depending on size of each fruit) fruits were picked from different locations of the sample box to constitute 25 g of sample. Twenty five grams of produce sample were transferred into 225 ml of appropriate enrichment broth such as lauryl sulfate tryptose (LST, Neogen, Inc., Lansing, Michigan), modified Escherichia coli (EC, Neogen, Inc.) broth, and universal pre-enrichment broth (UPB, 5 g tryptone, 5 g proteose peptone, 15 g KH2PO4, 7 g Na2HPO4, 5 g NaCl, 0.5 g glucose, 0.25 g MgSO4.7H2O, 0.1 g ferric ammonium citrate, and 0.2 g sodium pyruvate per liter) in sterile stomacher bags. The sample was mixed with the enrichment broth in a stomacher (Tekmar Co., Cincinnati, Ohio) for 2 min.
Coliform and Escherichia coli counts were determined by the three-tube most-probable-number (MPN) system using three 10-fold dilutions in 9-ml tubes of LST (Neogen, Inc., Lansing, Michigan) broth. The LST tubes were incubated for 24 and 48 h at 37C. LST tubes showing growth and gas production were transferred to 9-ml brilliant green bile (BGB; Neogen, Inc.) broth tubes each containing a Durham’s tube. The BGB-broth tubes were incubated at 37C for 24 and 48 h for selective enrichment for coliforms. Broth cultures from the tubes showing both growth and gas production were streaked on eosin methylene blue (EMB; Neogen, Inc.) plates. The EMB plates were incubated for 24 h at 37C, and the plates were then examined for characteristic E. coli colonies, that had a dark center, with or without a greenish metallic sheen. The suspected E. coli colonies were confirmed using indole, methyl red, Voges Proskauer, and citrate fermentation tests. Predominant coliform types were determined by identifying the isolated colonies from highest dilution of the sample on EMB plates, using Analytical Profile Index (API 20E) strips (bioMerieux, Marty l’Etoile, France). The detection limit for this method was 10 CFU g-1.
Analysis for Escherichia coli O157:H7 detection started by blending 25 g of sample in 225 ml of EC broth (Neogen, Inc.) supplemented with novobiocin (ICN Biomedicals, Inc., Irvine, California) by adding 2.25ml of 2 g L-1 novobiocin stock solution. This enrichment was incubated without shaking at 35C for 24 h (11). After the incubation 1 ml of the enrichment was mixed with 20 l suspension of magnetic beads (Dynal ASA, Oslo, Norway) coated with anti-O157 antibody. This mixture was incubated at room temperature for 30 min with gentle shaking. Tubes containing the mixture of culture and magnetic beads were placed in a strong magnet (Miltenyi Biotech, Inc., Auburn, California) for 5 min. After the separation of the beads, the liquid was discarded. The beads were then resuspended in 1 ml of buffered peptone water (BPW, 10 g peptone, 5 g NaCl, 3.5 g Na2HPO4, 1.5 g NaH2PO4 per liter) that contained 0.05% Tween 20 (ICN Biomedicals, Inc.),, incubated with gentle shaking for 5 min at room temperature, and the tubes were placed back into the magnet for 5 min. The liquid was discarded, making sure that no beads were lost. This washing step was repeated two more times. After the final wash, the beads were resuspended in 100 l of BPW, and plated on sorbitol MacConkey agar (Neogen, Inc.) supplemented with 2.5 mg L-1 potassium tellurite and 0.05 mg L-1 cefixime (Sigma, St. Louis, Missouri). Colorless or pale colonies were tested for O157 antigen, using an E. coli O157:H7 latex-agglutination test kit (Oxoid, Ltd., Hampshire, U K). Escherichia coli O157:H7 ATCC 43895 was used as positive control, and it was determined that the analytical method of detection used in this study could detect 10 cells of E. coli O157:H7 per 25 g of lettuce (data not shown).
Detection of Salmonella was done using the standard official technique described in the Food and Drug Administration’s Bacteriological Analytical Manual (1) with minor modification. Twenty five grams of produce sample was blended with 225 ml of UPB (5 g tryptone, 5 g proteose peptone, 15 g KH2PO4, 7 g Na2HPO4, 5 g NaCl, 0.5 g glucose, 0.25 g MgSO4.7H2O, 0.1 g ferric ammonium citrate, and 0.2 g sodium pyruvate per liter), and the enrichment was incubated for 18 to 24 h at 35C (19). The pH of enrichment for acidic fruits such as strawberry and raspberry didn’t reach below 6.0. One milliliter of the pre-enriched sample was transferred to each of 9 ml of tetrathionate (TT) and Rappaport Vassiliadis (RV) broths (Difco, Becton Dickinson, Sparks, Maryland). The TT broth tubes were incubated at 37C, and the RV broth tubes at 42.5C for 24 h. Broth in the tubes showing distinct turbidity were then streaked on xylose lysine desoxycholate (Neogen, Inc.) and bismuth sulfite (Difco) plates using an inoculation loop of 25 µL, and the plates were incubated at 37C for 18 to 24 h. Suspected colonies were transferred onto lysine iron agar (LIA) and triple sugar iron (TSI) slants (Difco) by streaking and stabbing. The LIA and TSI tubes were incubated for 24 h at 35C, and suspected colonies that produced the characteristic biochemical reactions were confirmed using Sallmonella-Tek™ ELISA Test System (bioMerieux, Inc.). Salmonella serovar Typhimurium strain ATCC 14028 was used as the positive control and it was determined that this analytical procedure used in this study could detect as few as 10 cells per 25 g lettuce (data not shown).
(d) Pulsed field gel electrophoresis of the E. coli isolates:
PulseNet’s standardized laboratory protocol was used with minor modifications as previously described (2). Bacterial strains were grown overnight on TSA plates at 37°C. Bacterial colonies were suspended in cell suspension buffer (100 mM Tris; 100 mM EDTA, pH 8.0) and adjusted to an optical density of 1.3 to 1.4 using a spectrophotometer (Bio-Rad Laboratories, Hercules, CA). The 400 µl adjusted cell suspension was mixed with 20 µl of proteinase K (20 mg/ml stock) and an equal volume (400 µl) of melted 1% SeaKem Gold agarose (BioWhittaker, Rockland, ME) containing 1% sodium dodecyl sulfate. The mixture was carefully dispensed into appropriate wells of a reusable plug mold (Bio-Rad Laboratories, Hercules, CA). After solidification, the plugs were transferred to each of round bottom tubes containing 1.5 ml of cell lysis buffer (50 mM Tris HCl; 50 mM EDTA, pH 8.0; 1% sarcosine) and 0.5 mg/ml of proteinase K. Cells were lysed in a 54°C water bath for 2 h with constant and vigorous agitation at 175-200 rpm. After lysis, the plugs were washed twice with pre-heated water and four times with pre-heated TE buffer (10 mM Tris, 1 mM EDTA; pH 8.0) for 10-15 min per wash at 50°C with the same agitation. Plugs were stored in 2 ml of TE buffer at 4°C until ready for restriction digestion of DNA. DNA in agarose plugs was digested with 50 U/sample of XbaI (Promega, Madison,WI) for at least 2 h in a 37°C water bath. The plugs were loaded into the wells in a 1.4 % pulse field certified agarose gel. DNA restriction fragments were separated with a CHEF-DR III (Bio-Rad Laboratories, Hercules, CA) with pulse times of 5 to 50 seconds at 14°C for 19.5 h in 0.5× TBE buffer. The gel was stained with ethidium bromide (Sigma-Aldrich, St. Louis, MO), and restriction fragment patterns were photographed with a UV transilluminator. E. coli O157:H7 ATCC 43895 strain was used as the reference in this analysis. The dendrograms were generated for a subset of the E. coli isolates at the Minnesota Department of Health using BioNumerics from Applied Maths (www.applied-maths.com). For the rest of the E. coli isolates from pre-harvest produce, the gel-photos were visually analyzed using the method described by Tenover et al. (39).
(d) Statistical analyses
The average coliform and E. coli counts were recorded in logarithmic scale, and the average counts were compared between various produce types and between the three different farm types using Student’s t test (27). Percent prevalence of E. coli between various produce types, and between produce samples collected from the three farm types were compared using standard and multiple chi-square tests (28). The difference in counts and prevalence was considered statistically significant at a P-value of less than 0.05. Escherichia coli prevalence was calculated by using the number of samples tested positive for E. coli, and dividing that by total number of samples. The prevalence were compared at first individually using 2×2 Chi-square analyses (27), and then F-test for multiple Chi-square analyses was performed using the PC SAS system for Windows operating system (SAS version 9.1, SAS Institute Inc., Cary, NC). Longitudinal data analyses were performed using the Generalized Estimating Equation, for determination of time-trend in the E. coli prevalence (41). After performing an Exploratory Data Analyses (EDA), the independent correlation structure was used in the generalized estimating equation (GEE) model fit to the E. coli prevalence data. The GEE model was chosen using a backward elimination procedure. The chosen distribution was Binomial, as the outcome variable took only two values – 0 for absence and 1 for presence of detectable E. coli in a produce sample. The odds ratios, 95% confidence intervals and the P- values were obtained from the PC SAS output. The Chi-square-values, F-values and the odds ratios (OR) were considered statistically significant at P < 0.05.
The activities leading to this objective were completed and for the most part this objective was accomplished as proposed. The organic, semi-organic and conventional farms that took part during this 2-year study were located in the upper Midwestern states of Minnesota and Wisconsin. A total of 24 semi-organic farmers participated in each of 2003 and 2004 harvest seasons by allowing sampling of fruits and vegetables from their farms, directly from fields (Table 1). Fourteen organic and 19 conventional farms participated in 2003, while the numbers of organic and conventional farms participating in 2004 were 8 and 14, respectively. In the year 2004, several organic farms in the area decided not to renew their certification documents, and some stopped growing produce. These uncontrollable circumstances reduced the participation of organic farms in 2004, compared to 2003.
A total of 2,029 samples of fresh fruits and vegetables were collected and analyzed in 2003 and 2004 (Table 1). In 2004, 335 more samples were collected as compared to those tested in 2003, representing a 39% increase. The average number of produce samples per farm was 16 for the semi-organic and conventional types and 13 for organic farms in 2003. In 2004, the average number of samples per farm was 22, 37 and 25 for the semi-organic, organic and conventional types, respectively. The four major produce types that contributed to approximately 50% of the total number of samples were leafy greens, cabbage, peppers and tomatoes. Other major types of produce were berries, cucumber, zucchini, lettuce, summer squash, broccoli, bok-choi and cantaloupe. Berries included strawberries, raspberries and blueberries. Other produce types collected and analyzed in smaller numbers were melon, kohlrabi, green beans, sprouts and peas. The produce types that had the greatest contribution increasing the number of samples in 2004 were leafy greens (62% more), peppers (74% more), tomatoes (42% more) and berries (81% more). The produce types that had the greatest contribution to the increase in the number of samples in 2004 by farm type were organic leafy greens (47 more samples), semi-organic berries (46 more samples), organic peppers (32 more samples), semi-organic leafy greens (30 more samples), semi-organic tomatoes (29 more samples) and organic tomatoes (26 more samples).
The percentages of each of the different produce types in the total sample lot varied between different produce types, and between the three different farm types (Table 2.2). During the two sampling seasons leafy greens was the produce type that accounted for more than 16% of the organic and semi-organic samples, but it was never greater than 4% of the conventional ones. Peppers were the produce type that contributed more to the conventional samples with more than 15%. More cucumber and zucchini samples were collected from conventional than from semi-organic and organic farms which resulted in significantly greater percentage distribution.
Coliform bacteria were detected in approximately 70% of the semi-organic fruits and vegetables in each of the two years of sampling. Among organic produce, 84% in 2003 and 80% in 2004 were coliform-positive samples. For conventional produce, 75% in 2003 and 64% in 2004 were positive for the presence of coliforms. In the two years of this study, the average coliform counts for any of the types of farm ranged from 1.5 to 2.3 log MPN/g (Table 3). Conventional produce had significantly less coliform levels than organic and semi-organic produce in 2004, but in 2003 this difference was only significant for semi-organic samples. From a subset of 826 samples positive for coliforms, Enterobacter spp. was detected in 58% and Klebsiella spp. was identified in 28%. Enterobacter cloacae and K. oxytoca were the two predominant coliform species.
Among the produce samples that had detectable coliforms, the average counts ranged from 1.4 log MPN/g among semi-organic and organic berries in 2003 to 4.8 log MPN/g in semi-organic summer squash in 2004 (Table 4). The average coliform loads in leafy greens, cucumber and tomato did not differ significantly between different farm types or between two years of sampling. In general, produce types such as pepper, tomato and berries had a lower range of average coliform loads from 1.4 to 2.7 log MPN/g, compared to produce types such as leafy greens, lettuce, cabbage, summer squash, zucchini, and cucumber, which had a range of 2.2 to 4.8 log MPN/g. In case of any of the nine major produce types, the average coliform population did not show any particular trend between the two years of sampling. For the nine major produce types, conventionally grown ones had either significantly lower or similar average coliform counts compared to their semi-organic and organic counterparts. However, conventional berries in 2004 had significantly greater average coliform population, compared to that in semi-organically grown peppers and berries.
The majority of the produce samples did not have detectable levels of Escherichia coli contamination. In the two years of sampling, 68 (8%) of semi-organic samples and 34 (7%) of organic fruits and vegetables had detectable levels of E. coli contamination at farms. For conventional produce, as many as 13 (2%) samples tested positive for E. coli. Among the 24 participating semi-organic farms, 11 farms in 2003 and 14 in 2004 had at least one E. coli contaminated produce samples (Fig. 1A, 1B). Four of the 14 organic farms in 2003 and 7 of the 8 organic farms in 2004 had at least one E. coli contaminated sample. Among the conventional growers, 3 out of 19 in 2003 and 5 of the 14 in 2004 had at least one E. coli-positive produce sample.
The prevalence of E. coli in those farms, which had at least one contaminated produce sample, ranged from 3.7 to 50% in 18 of the 57 participating farms in 2003, and from 1.4 to 50% in 26 of the 46 participating farms in 2004 (Fig. 1). In the year 2003, the E. coli prevalence in semi-organic farms varied from 4.8% to 50%, while the prevalence in organic farms was within a narrower range of 11% to 33% (Fig. 1A). The prevalence of E. coli in conventional farms was never greater than 7% in 2003. In 2004, E. coli prevalence in semi-organic farms ranged from 2 to 50%, while the prevalence in organic farms varied from 1 to 24% (Fig.1B). The prevalence in conventional farms in 2004 varied within a wider range of 3 to 38%, compared to the prevalence range among this farm type in 2003.
Out of the 18 semi-organic farms that had at least one E. coli contaminated produce sample, 7 had such contamination in both 2003 and 2004 and these are farm #1, 2, 5, 6, 7, 8, and 9. Among these, farm #5 had 50% E. coli prevalence in both years. Farm #8 and #9 had consistent prevalence of approximately 11 to 12% and 8 to 9%, respectively, in both the years. Among the organic farms, 7 had at least one produce sample that tested positive for E. coli. Four of these organic farms, #31, 33, 37, and 42, had these indicator bacteria in their produce in both years. The E. coli prevalence in these four organic farms decreased by 1.4 to 15-fold from 2003 to 2004. Seven conventional farms had at least one E. coli-positive produce sample, and farms #47 and #50 had such contamination in both years of this study. In farm #47, the E. coli prevalence increased 10-fold from 2003 to 2004, while that in the farm #50 remained approximately the same.
The prevalence of E. coli in leafy green samples from semi-organic and conventional farms was as much as 3-fold greater than that in organic samples. However, only the difference between the prevalence in semi-organic leafy greens in 2003 was significantly greater than that in the organic ones (Table 5). When the prevalence for leafy greens was compared among the three farm types for both years combined, the E. coli prevalence in semi-organic (18%) and conventional (24%) leafy greens had significantly greater than the organic (8%) counterparts (P < 0.05). The E. coli prevalence in lettuce ranged from 0% in conventional samples in 2003 to 25% in the same type of farms for 2004 (Table 5). The prevalence in semi-organic cabbages was approximately 2 to 4-times the prevalence in conventional cabbages, for both years. However, these differences were not statistically significant. Organic cucumbers had 8 to 12% E. coli prevalence, which was approximately 2 to 4-times the prevalence in semi-organic cucumbers. This difference in prevalence was not statistically significant. None of the semi-organic, organic and conventional tomato, cantaloupe, apple, and kohlrabi samples had E. coli contamination, in either 2003 or 2004.
Leafy greens and lettuce samples had the greatest prevalence of Escherichia coli than any other type of produce when all the three farm types were combined for both years (Fig. 2). Cabbages had significantly lower E. coli prevalence compared to that in lettuce and leafy greens in 2003, and significantly greater prevalence compared to that in pepper and berries. When zucchini, summer squash and cucumber were clustered together as a single produce category, the prevalence of E. coli was similar to the one observed in peppers and berries. However, that prevalence was significantly lower than that in lettuce, leafy greens and cabbages. Within each category of produce, the E. coli prevalence was not significantly different between the two years of the study.
Fresh produce that were contaminated with E. coli had approximately 2.0 to 2.4 log MPN/g (Table 2.3). Both semi-organic and organic produce samples, had wide ranges of 1.0 to 6.3 and 0.6 to 7.8 log MPN/g of E. coli, respectively. The range of E. coli counts in conventional produce samples had a narrower range of 1.4 to 4.0 log MPN/g counts. Among various major produce types, lettuce, leafy greens and cabbage had 2.2 to 2.4 log MPN/g of E. coli counts. All other produce types that had E. coli contamination had an average count of 1.9 log MPN/g. However, these differences in counts of the indicator bacteria between different produce types or between farms types were not statistically significant.
When the E. coli prevalence in different produce types were compared between the three different farm types, only leafy greens had significantly different prevalence between organic and semi-organic farms (Table 5). Thus the major trend in the E. coli prevalence in the present study was observed between various produce types (Fig. 2). Visible presence of soil particles in between the leaves in lettuce, leafy greens and cabbages might have caused significantly greater E. coli prevalence in these produce types compared to the other varieties such as zucchini, summer squash, cucumber, pepper and berries.
All of the 2,029 fruits and vegetable samples collected during 2003 and 2004 seasons were tested for the presence of Salmonella and Escherichia coli O157:H7. Neither of these two pathogenic bacteria was detected from any of the produce samples.
This study confirmed our previous finding that organic fruits and vegetables do not appear to be more susceptible to pre-harvest contamination by Salmonella and E. coli O157:H7 compared to conventional produce. We observed that for some of the fruit and vegetable types, conventional produce had significantly lower coliform levels than semi-organic and organic vegetables. However, the type of farm operation appeared to have little influence on the prevalence of contamination with E. coli by produce type. The presence of E. coli in produce had almost no difference between the two years of the study. Produce types such as leafy greens, lettuces and cabbages appeared to be more susceptible for E. coli contamination, for the three types of farm in both years. This finding suggests that if foodborne pathogens were present under similar environmental conditions, leafy vegetables may be a likely vehicle of transmission.
In the spring and fall seasons, samples of soil, manure, compost and water were collected in selected farms that had vegetables positive for E. coli, and were subjected to microbiological analysis. From a total of 80 samples, 8 were positive for this bacterium. Using DNA fingerprinting analysis, we were able to determine that none of these environmental isolates were related to any of the strains obtained from produce.
Ten organic and 19 semi-organic farms participated in this part of the study by providing samples of fresh fruits and vegetables, collected directly from the farm fields. Out of the 19 semi-organic and 10 organic farms, 17 and 6 provided samples in each of the harvest months, respectively. Two and three semi-organic and organic farms provided samples during three months. During each of the harvest months, produce samples were collected from 21 to 29 organic and semi-organic farms in 2003 and 2004 (Table 6). Only 2 to 3 of these farms provided samples during the four harvest months in the first year of this study (Table 6).
A total of 1,432 (888 semi-organic and 544 organic samples) pre-harvest produce samples were collected and analyzed during the three-year study (Table 7). Approximately a dozen different produce types that included lettuce, spinach, kale, collards, mixed greens, cabbage, broccoli, pepper, cucumber, zucchini, summer squash, pepper, tomato and berries were analyzed. Among the four months of harvest for the three seasons and two farm types combined, the number of produce samples ranged from 192 in September to 562 in August (Table 7). Seventy two organic produce were collected in September of 2002, 2003 and 2004, while a total of 235 were collected in August. The numbers of semi-organic produce samples ranged from 120 during September to 327 in August for both years combined. No organic produce was available for collection in September of 2002 and only 4 such produce samples were collected in September of 2003.
On average, the coliform contamination in fresh produce samples was approximately 2.5 to 3.5 log MPN g-1 (Table 8). The coliform count in organic produce was as low as 2.2 log MPN g-1 in June and this count was significantly lower compared to the average coliform population in organic produce collected during July and August. Among semi-organic produce, the average coliform count in the samples collected during September was 2.3 log MPN g-1. This count was significantly lower compared to the average coliform population in produce collected during June and July. Overall, the average coliform population was 3.0 log MPN g-1 for both semi-organic and organic fruits and vegetables. The average Escherichia coli counts in pre-harvest produce that tested positive varied from 1.4 to 2.4 log MPN g-1 (Table 8). The average counts of this fecal indicator bacterium in semi-organic produce decreased from 2.2 log MPN g-1 in June to 1.4 log MPN g-1 in September (P = 0.08). In organically produced fruits and vegetables, however, the average E. coli counts increased from 1.7 log MPN g-1 in June to 2.4 log MPN g-1 in July, and to 2.3 log MPN g-1 in August (P = 0.1). However, none of the organic produce collected during September had detectable E. coli contamination. These variations in the average E. coli population during the four months of the three harvest seasons were not significantly different for semi-organic and organic produce.
On an average, the percentage of leafy vegetables, which included lettuce, spinach, kale, collards, mixed greens, and cabbage varied over the four months of June, July, August, and September (Fig. 3). In the month of June, on an average, 70 to 80% of the samples from semi-organic and organic farms were leafy vegetables. This average percentage decreased to approximately 55% during July. During August and September, the average percentage of leafy vegetables from these farms was only 10 to 30%. The average percentage of E. coli positives in produce from semi-organic farms was 6 – 25% during June and July (Table 9). This percentage decreased to 1.3 – 2.4% during August and to non-detectable (ND) levels during September. Figure 4.2A and 4.2B showed the plots of the sample-level E. coli prevalence trends over the four harvest months, separately for 2003 and 2004. In 2003, 12.5% of the organic produce had detectable E. coli contamination and as many as 27.5% of the semi-organic produce tested positive for such contamination. In August 2003, approximately 12% of organic fruits and vegetables were contaminated with this fecal indicator bacterium, but none of the semi-organic produce had such contamination. In 2004, the E. coli prevalence 4 to 5% in June, 15 to 17% in July, 2 to 4% in August, and reached below 2% prevalence in September, for both organic and semi-organic fruits and vegetables.
Overall, across the three years of sampling, the months June (OR = 15.6, 95% CI = 2.06 – 118.11, P = 0.008) and July (OR = 18.2, 95% CI = 2.46 – 135.15, P = 0.004) significantly increased the risk of E. coli contamination in pre-harvest produce, compared to the months of August and September (Table 9). Leafy vegetables such as lettuce, spinach, kale, cabbage had a 2.4-times greater risk of E. coli contamination compared to any other fruit and vegetable types (OR = 2.4, 95% CI = 1.42 – 4.09, P = 0.001). Among the two farm types, none appeared to be more susceptible to E. coli contamination compared to the other. When the prevalence of E. coli was evaluated for the year 2004 separately, both organic and semi-organic produce had approximately 5%, 16%, 3% and 1% E. coli prevalence in June, July, August and September, respectively (Fig. 4B). Such consistency in percentage of E. coli positives was not evident for produce collected in 2003 (Fig. 4A). Including all the fruits and vegetables from the two farm types together, the risk of E. coli contamination was significantly greater in June (OR = 4.7, 95% CI = 1.48 – 34.09, P = 0.008) in 2003 and in July (OR = 22.9, 95% CI = 3.06 – 172.00, P = 0.002) in 2004. Leafy vegetables were consistently at a greater risk of E. coli contamination in both 2003 (OR = 2.8, 95% CI = 1.34 – 5.23, P = 0.004) and 2004 (OR = 2.4, 95% CI = 1.24 – 4.6, P = 0.009).
The risk of E. coli contamination in pre-harvest produce was no different between organic and semi-organic farms. However, the interaction term between farm types and months of sampling showed that the organic produce were at a significantly greater risk of contamination from this fecal indicator bacteria compared to the semi-organic ones in June (OR = 21.1, 95% CI = 4.71 – 156.42, P < 0.0001) and July (OR = 21.7, 95% CI = 5.12 – 162.04, P < 0.0001) (Table 10). The risk of E. coli contamination was significantly greater in leafy vegetables compared to any other produce types, also during June (OR = 21.1, 95% CI = 4.65 – 160.32, P < 0.0001) and July (OR = 21.3, 95% CI = 5.45 – 158.60, P < 0.0001). Such significance was not evident in August and September. When the E. coli prevalence in different types of leafy vegetables such as lettuce, spinach, cabbage, and other types like kale, and Swiss chard was compared between the four different months, we found that E. coli prevalence was 2 to 4-times greater in each of these leafy vegetables during June and July, compared to August. However, none of these differences within each individual leafy vegetable type were statistically significant.
Variation in risk of E. coli contamination, separately for leafy vegetables and all other vegetable types is presented in Table 11. For both of these produce types, risks of E. coli contamination at farms were significantly greater in June (OR = 23.62, 95% CI = 4.18 – 111.70, P-value < 0.0001 for leafy vegetables, and OR = 24.92, 95% CI = 2.98 – 132.46, P-value < 0.0001 for all other vegetable types) and July (OR = 23.74, 95% CI = 3.52 – 121.27, P-value < 0.0001 for leafy vegetables, and OR = 25.86, 95% CI = 2.56 – 134.21, P-value < 0.0001 for all other vegetables) compared to August and September. The risks during the four harvest months did not differ between 2003 and 2004, for either leafy vegetables (OR = 1.31, 95% CI = 0.74 – 2.28, P-value = 0.3617) or other vegetable types (OR = 0.90, 95% CI = 0.37 – 2.23, P-value = 0.8263). Also, the trend of E. coli prevalence in these four months in the leafy vegetables (OR = 0.96, 95% CI = 0.54 – 1.68, P-value = 0.8759) and other vegetable types (OR = 1.81, 95% CI = 0.77 – 4.24, P-value = 0.1702) did not differ significantly between semi-organic and organic farm types. From a total of 100 E. coli isolates obtained from pre-harvest produce that were analyzed using PFGE, 27 of them had similarity values greater than 40%. A dendrogram obtained from the PFGE patterns of those 27 E. coli isolates showed 20 different band patterns (Fig. 5). The clusters A and C had identical PFGE patterns obtained from two pairs of E. coli strains isolated from two different produce, collected from the same farm one month apart. The cluster B had four identical PFGE patterns, and the E. coli strains AWK304, AWL203 and AWPE103 were from three different produce types collected from the same organic farm in 2003 and 2004. Overall, the dendrogram analysis indicated a large diversity of E. coli isolates obtained from produce samples from different farms. Most of the isolates obtained from each farm in the same and different year, had distinctly different PFGE patterns (Table 13). Objective 3)
Since there were no samples positive for the pathogenic bacteria, this objective was focused on identify management practices linked to E. coli prevalence as an indicator of fecal contamination. In 2004, approximately 88% of organic and 79% of semi-organic and all the conventional farms responded to the questionnaire on their farm management practices (Table 14). In 2003, these percentages ranged from approximately 79% among semi- organic and organic growers to 84% among conventional growers. Approximately 44 to 50% of conventional farms used animal waste as fertilizer, while 70 to 100% of the semi-organic and organic farms had animal manure as fertilizer. Among the organic farms, which used animal manure for fertilization of their produce plants, 90 to 100% used composting. Among semi-organic farms, this percentage ranged from 64 to 71 during 2003 and 2004. When farmers were asked about the duration of ageing of manure to be applied on produce plots as fertilizer, approximately 30 to 40% of organic and 40 to 50% of semi-organic farms used manure that was aged for more than 6 months. Among conventional growers that used animal waste for fertilization of their produce plants, as many as 90 to 98% used more than six months of ageing before application as fertilizer.
Use of animal wastes in fertilization of fresh fruits and vegetable plots significantly increased the risk of E. coli contamination in fresh produce grown in semi-organic (OR = 12.9, 95% CI = 2.9 – 56.3, P-value < 0.0001) and organic (OR = 13.2, 95% CI = 2.2 – 61.2, P-value < 0.0001) farms (Table 15). Although composting did not affect the E. coli prevalence in semi-organic and organic produce, ageing of non-composted manure for more than 6 months contributed to a significant reduction of the risk (OR = 4.2 95% CI = 1.7 – 12.3, P-value = 0.005) among organically produced fruits and vegetables. Semi-organic and organic farms used various types of animal wastes such as chicken, hog, horse and cattle manure. Among these different manure types, use of cattle manure for fertilization of produce crops increased the risk of E. coli contamination by approximately 7-fold (OR = 7.4, 95% CI = 1.6 – 36.8, P-value = 0.003). Farmers who used animal wastes for fertilization applied manure either in the Spring or in the Fall or both in Spring and Fall. However, these different times of application of manure application did not have any significant effect on the E. coli prevalence in semi-organic and organic produce.
Our research findings, both during the present study and in a previous study, showed that only 1.5 to 2.5% of pre-harvest conventional produce had detectable E. coli contamination (29, 30). This low number of positive samples prevented any risk-factor analyses involving conventional farms. However, among the 6 conventional farms which had at least one E. coli contaminated sample, 4 used manure-based fertilizers. In this study, only semi-organic and organic farms were included in the analyses of farm-level risk-factors associated with greater prevalence of fecal indicator bacterium. Among these two farm types, users of animal waste as fertilizer were at a significantly greater risk of possible fecal contamination in their produce compared to those who did not use manure-fertilizer (Table 15). A couple of recent reports have documented that application of untreated animal wastes such as hog manure and bovine manure into agricultural soils might lead to survival of E. coli in the soil for 70 to 170 days (17). However, these two studies reported conflicted findings on detection of the fecal indicator bacterium from vegetables grown in the manure-amended soils. Ingham et al. (13) reported persistence of E. coli contamination in carrot, lettuce and radish grown in the fertilized soil for more than 200 days. But Cote and Quessy (6) could not detect the bacterium from cucumbers, grown in the soil that was amended with liquid hog manure.
Ageing of animal wastes before their application as fertilizer for produce cultivation, was used by several of the participating farmers. Ageing of animal manure for less than 6 months increased the risk of fecal contamination by more than four folds in organic produce (Table 15). However, this risk-factor did not show a significant effect on E. coli prevalence in semi-organic produce. In a recent report, persistence of E. coli was monitored in agricultural soils in Wisconsin, amended with non-composted bovine manure. No clear effect of fertilization-to-harvest intervals of 90, 100, 110 and 120 days was reported on E. coli contamination in vegetables produced in the manure-amended soils (17). We found no strong association of fertilization-to-harvest interval of less than 90 days with increased risk of E. coli contamination in semi-organic and organic produce. Among the semi-organic and organic produce, using cattle manure for fertilization increased the risk of fecal contamination by 2-fold and 7-fold, respectively (Table 15). This finding was consistent with our previous report (29). In that study, organic produce cultivated using cattle manure in fertilizer showed 16% E. coli prevalence, compared to 7% in those which were cultivated using other type of animal manure-based fertilizer.
Farmers participating in this two-year study were grouped into four geographical locations such as Middle (M), Twin Cities (TC), South East (SE) and South (S) in Minnesota (Fig. 6), and into two distinct groups such as North (N) and South (S1) in Wisconsin (Fig. 7). The average percent prevalence of Escherichia coli in produce from semi-organic and organic farms in South East (SE), South (S), and Middle (M) portion of the state of Minnesota was significantly greater compared to that in farms located in the Twin Cities (TC) location (Table 16). In Wisconsin, organic and semi-organic produce from farms located in the South (S1) cluster had significantly greater average E. coli prevalence of 11% compared to 6% average prevalence in the farms located in the North (N) cluster. The distribution of leafy vegetables including lettuce and cabbage among these different locations in Minnesota and Wisconsin was quite consistent (Table 16).
The risk analysis on the basis of these regions of Minnesota and Wisconsin determined that produce from the South East (SE) region of Minnesota had approximately 3.5-times (OR = 3.45, 95% CI = 1.8 – 35.2, P = 0.008) greater risk of E. coli contamination compared to those from the South (S) region (Table 17). In Wisconsin, the organic and semi-organic produce from the farms located in the South (S1) region were at a 2.7-times (OR = 2.67, 95% CI = 1.3 – 9.4, P = 0.004) greater risk of such contamination compared to their counterparts cultivated in farms in the North (N) region. All the participating conventional produce growers were from Minnesota, and 6 of the 19 of these farmers had at least one E. coli contaminated samples from their farms. Prevalence of this fecal indicator bacterium in conventional produce from Middle (M), and South (S) regions of the state were 4.4 and 6.9%, respectively. None of the conventional produce grown in the Twin Cities (TC) and South Eastern (SE) region of Minnesota had detectable E. coli contamination.
We communicated directly with each of the participating farmers reporting their individual results and the overall season results after each season. Unfortunately, we were not very successful in attracting the participating farmers to a meeting.
To the scientific community, we presented results of the different parts of this project at the World Congress of Organic Foods, at the Global Good Agricultural Practices Conference in early 2005 and at the International Association for Food Protection Annual meeting in 2005 and 2006.
After a previous study similar to this project was published in 2004, our work received considerable attention from the media and specially form organic and sustainable agriculture websites. We expect that once the results of this project are made public, there will be great interest from the agricultural, scientific and consumer communities.
No Information Reported
No Information Reported
No Information Reported
Educational & Outreach Activities
1. Mukherjee, A., D. Speh, A. T. Jones, K. M.Buesing, L. T. L. Xiong, and F. Diez-Gonzalez. 2006. Longitudinal microbiological survey of fresh produce grown by farmers in the Upper Midwest. J. Food Prot. 69:1928-1936.
2. Mukherjee, A., D. Speh, and F. Diez-Gonzalez. 2006. Association of Farm Management Practices with Risk of Escherichia coli Contamination in Pre-harvest Produce Grown in Minnesota and Wisconsin. Int. J. Food Microbiol. (submitted).
1. Mukherjee, A., D. Speh, A. Jones, L. Xiong, K. Buesing, and F. Diez-Gonzalez. 2005. Microbial safety evaluation of organic and conventional fresh produce at the pre-harvest stage. IAFP Annual Meeting, August 14-17, Baltimore, MD.
2. Diez-Gonzalez, F. 2005. Manure application and harvest intervals. Global Good Agricultural Practices Conference, Orlando, FL, January 12.
3. Mukherjee, A., D. Speh, and F. Diez-Gonzalez. 2004. Preharvest evaluation of the microbiological quality of organic and conventional vegetables. First World Congress in Organic Foods, Michigan State University, March 30.
4. Diez-Gonzalez, F. 2004. Control of Escherichia coli O157:H7 in cattle manure, First World Congress in Organic Foods, Michigan State University, March 30.
5. Mukherjee, A., D. Speh, E. A. Dyck, and F. Diez-Gonzalez. 2003. Assessment of the microbiological quality of organic fruits and vegetables at the farm level, IFT Annual Meeting, July 12-16, Chicago, IL.
6. Presentation on Food Safety at the Organic Field Day in SWROC, Lamberton, MN, April 26, 2003.
Areas needing additional study
This project provided many potential avenues of future research in the area of microbiological quality and safety of fresh fruits and vegetables. Neither E. coli O157:H7 nor Salmonella was detected in any of the produce samples. A study could look into the possibility of antagonistic effect of native microbial population on these foodborne pathogens. A study could be designed to look into seasonal variation of contaminants in fresh produce across a longer time-scale, as they are being cultivated. Such longitudinal study could also look into seasonal factors such as temperature and moisture content in air and soil in the environment in which produce were grown. Similar surveys in states, where produce industry is a substantial part of their agro-economy, could provide important insights in improving quality and safety of their fresh produce.
Because of the increased contamination susceptibility of lettuce and leafy greens, a more balanced comparison of organic and conventional lettuce and leafy greens could be an ideal study to obtain an assessment whether organic produce are more or less safe than its conventional counterparts. In addition to this potential project, it is critical to determine the actual source of E. coli contamination which can be done by means of controlled experiments of trace-forward and trace-back analysis. This approach could help to answer the question whether E. coli can become adapted to grow in the soil. The reduction of the E. coli contamination as affected by time is an intriguing result, which could be due to the die-off of this organism as time passes or could be due to a shift of wild animal population.
Finally, this study stresses the need of devoting resources to investigate the soundness and validation of the current organic recommendations regarding use of manure and compost. The survival of foodborne pathogens in soil, produce and the environment is an area that needs further attention that would help to establish recommendations that would lead to safer organic products.