The rate of decrease in the E. coli level observed in manure-fertilized soils in this study was within the range of bacterial death rates in soil or manure-fertilized soils reported elsewhere for microcosm and field studies. Guo et al. (23) observed a 1- to 1.5-log decrease in the levels of salmonellae in inoculated topsoil during 45 days of 20°C storage in greenhouse trays. The levels of a nonpathogenic E. coli strain decreased by 6 log CFU/g in 60 days at 4 and 20°C and in 12 days at 37°C in nonsterile sieved soil in flasks (4). Recorbet et al. (42) reported that E. coli death in inoculated soil microcosms stored at 28°C was dose dependent, with 3- to 4-log CFU/g decreases occurring within 5 days. In a field study involving application of contaminated pig slurry (feces plus urine) to soil, Baloda et al. (2) detected Salmonella enterica serovar Typhimurium DT12 in treated soil 14 days after manure application but not 21 days after application. Stoddard et al. (51) reported a 6.9-day half-life for fecal coliforms in bovine manure-fertilized soil, although the rate of decline decreased with time. Van Donsel et al. (59) reported that fecal coliform levels in manure-fertilized soils decreased by 3 log CFU/g in 2 to 4 weeks, depending on the ambient temperature and the exposure to direct sunlight. Similar trends were reported by Tannock and Smith (53) for salmonellae.
For all three manure-fertilized soils in this study, persistence of indigenous E. coli past the end of the study was common. Persistence of fecal bacteria in soil has been reported elsewhere. For example, Jones (30) described E. coli that survived for at least 60 days in soil at 25°C and for at least 100 days at 4°C. Bolton et al. (5) detected E. coli O157:H7 in soil 99 days after a fecal suspension containing this organism was applied to grassland. It is possible that fecal bacteria may enter a viable but nonculturable state in soil after an extended time (56), although it is uncertain whether pathogenic bacteria in this state would be infective. The persistence of E. coli in the present study was likely either the result of bird and mammal defecation or movement within the plots (tracks and droppings were observed at all three locations, with greater prevalence on the silt loam and silty clay loam soils) or the result of protection of bacteria within soil particles. In control plots to which manure had been applied the preceding fall (2002; loamy sand and silt loam), sporadic enrichment-positive results were obtained throughout the 2003 growing season. When no manure was applied in either the fall of 2002 or the spring of 2003 (silty clay loam), enrichment-negative results were obtained early in the spring of 2003 and throughout the 2003 growing season. However, these enrichment-negative results were interspersed with enrichment-positive results, suggesting that bird and/or mammal recontamination was the cause of the apparent persistence of indigenous E. coli in manure-fertilized soils. Together, the results of the manure-fertilized soil analyses suggested that application of noncomposted bovine manure far enough in advance of vegetable harvest would present little additional risk beyond that associated with nonfertilized soil. Furthermore, our observations and data suggested that bird and mammal activity may result in contamination of vegetable production soils regardless of when or if manure is used as fertilizer.
In our study, sporadic low-level E. coli contamination of vegetables occurred even when the NOP ≥120-day limit was followed. In light of previously published results for colonization of vegetables by pathogenic bacteria, these results are not surprising. In a study of vegetables grown in inoculated soil under greenhouse conditions, Van Renterghem et al. (60) found that Listeria monocytogenes, previously added to soil at a level of 5 log CFU/g, was detected by enrichment on three of six radish samples 3 months after inoculation. In contrast, none of six carrot samples tested positive and only one of six samples of radish soils tested positive. Although the numbers of samples were low, these results suggest that the contamination rates for different vegetables may vary under identical growing conditions and that vegetables may become contaminated with soilborne pathogens early in growth. Several studies have suggested that colonization of plants is most likely when the plant is a seedling, and the emerging root is a key area of bacterial attachment (11, 14). Subsequent internalization of soilborne pathogenic bacteria is also greatest when the plants are seedlings, and internalization is less likely as the plants mature (11). The ability to colonize vegetable roots and shoots varies for different strains of a bacterial species (3, 62) and for different bacterial species (3, 10, 14). Bacterial motility (11) and the ability to use seed exudates as carbon sources (45) are related to the extent of colonization. Internalization of bacteria in seedlings is more likely in hydroponic systems than in soil (24, 63), but internalization and transport throughout the plant can still occur in the latter growth medium (49). Collectively, the literature on bacterial colonization of plants and internalization suggests that the critical time for preventing vegetable contamination with manure fertilizer may be at the time of planting. A better alternative to an application-to-harvest interval may be an application-to-planting interval.
Enrichment-negative results were obtained more often for vegetables than for soil samples. The possible causes of this trend included recontamination of soil, but not vegetables, by birds and other animals, competitive exclusion of E. coli by other microbes on the vegetable surface, and release of antimicrobial plant-derived compounds when samples were stomached prior to enrichment.
Examination of our results leads to two seemingly contradictory conclusions about the current NOP 120-day application-to-harvest interval. One conclusion, based on the persistence of indigenous E. coli in the soil and the occasional detection of indigenous E. coli on carrots harvested >120 days after manure application, is that even the current 120-day limit does not completely guarantee the absence of contamination. The second conclusion is that decreasing the 120-day limit to 100 days would only slightly increase the risk of contamination. Although the prevalence and shedding of pathogenic E. coli in cattle may be affected to some extent by dietary changes (12, 13, 25, 46), it is apparent from studies of herds (17, 43, 55, 65) and slaughterhouses (44, 57) that carriage of E. coli O157:H7, other Shiga toxin-producing E. coli, and Salmonella spp. within herds is common. Furthermore, some research suggests that pathogenic E. coli can persist within a cattle farm for several years (48). Therefore, it is prudent to assume that any noncomposted bovine manure applied to vegetable production fields contains these pathogens. However, the sporadic pathogen carriage and shedding for most animals in a herd (22) should result in lower levels of pathogenic bacteria than of indigenous E. coli in bovine manure. Because indigenous E. coli has been shown to behave like the two main enteric bacteria that cause vegetable-associated outbreaks (39, 40), it could be argued that a large reduction in the level of indigenous E. coli in manure-fertilized soil that occurred within a ≥100-day application-to-harvest interval should result in a satisfactorily low likelihood of vegetable contamination by manure-borne E. coli O157:H7 and Salmonella spp. However, it is also possible that some pathogens more readily colonize vegetables and/or survive in manure-fertilized soils than indigenous E. coli does. In the present study, reductions in the indigenous E. coli levels of at least 3 log CFU/g of manure-fertilized soil were readily achieved within this application-to-harvest interval. For lettuce and carrots grown in the present study, the combination of a 100-day application-to-harvest interval and passive washing for 30 s under running tap water resulted in frequent enrichment-negative results when vegetables were tested for indigenous E. coli. The two-step enrichment procedure used to detect E. coli was very sensitive; a detection limit of approximately one cell was determined in supplemental, broth-based experiments (data not shown). However, the ability of pathogens to grow to detectable levels during this procedure was not tested. Given the sensitivity of the two-step enrichment method, the presumed greater numbers of indigenous E. coli than of enteric pathogenic bacteria in bovine manure, and the gentleness of the washing treatment, a 100-day application-to-harvest interval may result in a satisfactory reduction in the risk of pathogen contamination.
An appropriate manure application-to-harvest interval cannot be a stand-alone guarantee against vegetable contamination. Agricultural practices other than fertilization with manure can result in transfer of pathogenic bacteria from soil to vegetables. For example, Wachtel et al. described E. coli O157:H7 contamination of vegetable roots via soil after the soil was irrigated with contaminated water in laboratory (61) and field (62) studies. Irrigation with contaminated irrigation water may also lead to pathogen colonization of vegetable leaves (50). Furthermore, unclean conditions may contaminate vegetables after they are harvested (54). Therefore, use of a 100-day application-to-harvest interval would be contingent on thorough washing and strict adherence to good agricultural practices.
In conclusion, examination of our results suggests that using the current NOP manure application-to-harvest interval of ≥120 days does not guarantee the absence of manure-borne bacteria from vegetables grown in manure-fertilized Wisconsin soils. However, we concluded that decreasing the NOP manure application-to-harvest interval from ≥120 to ≥100 days for typical Wisconsin soils would only slightly increase the risk of vegetable contamination. Such a decrease would provide Wisconsin vegetable farmers with greater flexibility in scheduling manure application and would lessen the likelihood of soil compaction caused by manure application too early in the spring. Other agricultural factors, such as the application-to-planting interval, harvest hygiene, and postharvest washing treatments, are likely also critically important in preventing contamination of vegetables with pathogenic bacteria. The findings of the present study should tested in field studies with manure-borne pathogens.
This work was supported by a grant from the United States Department of Agriculture National Research Initiative.