Behaviour of the pathogen surrogates Listeria innocua and Clostridium sporogenes during production of parsley in fields fertilized with contaminated amendments

Abstract

The survival and transfer of Listeria innocua and Clostridium sporogenes, used as surrogates of the food borne pathogens Listeria monocytogenes and Clostridium botulinum, were quantitatively assessed under field conditions. In the soil, spores of C. sporogenes declined by less than 0.7 log cycles within 16 months and were detected on parsley leaves throughout the experiment. In contrast, L. innocua in the soil declined by 7 log cycles in 90 days and was detected on leaves in low numbers (>0.04 MPN g⁻¹) during the first 30 days. Rates of decline in soil were similar in the laboratory at 20°C for two strains of L. innocua and L. monocytogenes; and in the field for L. innocua over two different years. L. innocua survived better in winter, indicating an important influence of temperature. The major cause of transfer of L. innocua from soil to parsley leaves was splashing due to rain and irrigation. As few as 1 CFU g⁻¹Listeria in soil led to contamination of parsley leaves. Internalisation of Listeria through parsley roots was not observed. Under the conditions of soil and climate studied, a delay of 90 days between application of potentially contaminated fertilizer and harvest should be sufficient to eliminate L. monocytogenes.

Listeria, Clostridium, Soil, Compost, Fresh produce, Parsley

1 Introduction

In the EEC countries, the use of waste from animal farming (manure) and from water treatment processing (sewage sludge) in agriculture constitutes an economically acceptable alternative for the intensive use of chemical fertilizers and contributes to waste reduction. Hence, these practices are likely to increase in the future. However, the risk of contamination of crops with food borne bacterial pathogens that could be present in wastes should be fully evaluated.

The food borne pathogens Listeria monocytogenes and Clostridium botulinum are both ubiquitous bacteria widely distributed in agricultural environments, including soil and organic amendments and products [1–7]. Their occurrence in organic amendments can be frequent. The incidence of L. monocytogenes in organic fertilizers made from sewage sludge is as high as 60–73%[4]. A recent survey on bio-compost showed that more than 50% of the tested samples were contaminated with toxigenic C. botulinum strains (i.e., 36/64 positive samples) [8]. The use of organic amendments for soil fertilization may therefore introduce L. monocytogenes or C. botulinum populations in the crop environment which may subsequently be transferred to plant produce and consequently constitute a health hazard for consumers. Cabbage grown in fields fertilized with contaminated manure was at the origin of one of the most serious outbreaks of listeriosis [9] and C. botulinum has caused several outbreaks linked to both fresh and heat treated vegetables [10]. However, organic fertilizers are not applied to fields immediately before harvest, and therefore the risk for consumers depends on the ability of the introduced food borne pathogens to survive and persist in the crop environment.

The persistence of L. monocytogenes in soil has been studied by some authors, but most of the data have been obtained in soil microcosms, which usually consist of autoclaved soil, and might not reflect the behaviour of this pathogen under field conditions [5,11–14]. To date, data obtained under field conditions have concerned L. monocytogenes populations naturally present in organic amendments at very low densities that may not reflect high levels of contamination [15,16]. In addition, use of very low levels of inoculum does not permit quantitative analysis of survival over time and of transfer from soil to plants.

The aim of this study was: (i) to assess quantitatively the behaviour of C. botulinum and L. monocytogenes, (ii) to work under field conditions and not in laboratory microcosms. For this reason, the non-pathogenic surrogates, Listeria innocua and Clostridium sporogenes, were used and were introduced to soil through organic fertilizers. Their persistence in soil and their transfer to plants were assessed through cultivation periods over two and three successive years for C. sporogenes and L. innocua, respectively. The identity of strains recovered from the surrogate populations was carefully determined by colony typing. The predominant routes of transfer from soil to plants, (viz., splashing, air transmission and plant internalisation) and survival on plants were evaluated for L. innocua populations.

2 Material and methods

2.1 Bacterial strains

Strains of L. innocua (CIP 80–12) and C. sporogenes (CIP 79-3) were purchased from the Pasteur Institute Collection (Paris, France) and stored in 30% glycerol at −20°C. L. monocytogenes (EGDe) was kindly provided by Dr Hélène Bierne, Institut Pasteur, Paris, France. L. monocytogenes (LmP60) and L. innocua (LiP60) had been isolated in 2003 from organic amendments used on horticultural crops. Before inoculum production, strains were streaked on appropriate agar media to check purity.

2.2 Plant material and amendments

Seeds of parsley (Flat leave parsley Petroselinum sativum var. “Géant d'Italie”) were sown and raised in commercial substrate (TREF, Vulaines, France) in greenhouses for three weeks before being transplanted as three-leafed seedlings in experimental fields. Parsley was irrigated with groundwater using sprinkling ramps located in the middle of the experimental field, 1–3 times per week according to plant evapotranspiration data.

Two organic fertilizers were used: composted sewage sludge from a municipal waste water treatment plant in experiments conducted in 2000, 2001 and 2002, and composted bovine manure in 2003. Sewage sludge compost was provided by the Lyonnaise des Eaux (CIRCEE, Le Pecq sur Seine, France). Composted manure consisted of stable litter from a bovine farm, composted for 15 months, and was provided by the Rhône-Alpes Regional Experimental Unit of the CTIFL (SERAIL, Brindas, France).

2.3 Enumeration of bacterial populations

For soil and leaf samples, identical methods were used to enumerate population sizes of total mesophilic bacteria, Listeria and Clostridium. Samples were homogenized in phosphate buffer, pH 6.8, for plate counts or in selective enrichment broth for MPN counts (2.5 g of leaves and 4–5 g of soil in 20 ml). Total, culturable bacterial populations were determined by plating on respectively 1:10-diluted Tryptone Soya Agar (TSA), consisting of 3 g of TS Broth and 15 g l⁻¹ agar. Colonies were counted after 5–7 days of incubation at room temperature (ca. 25°C). Listeria populations were enumerated by direct plating on selective Oxford agar medium [17] or, for low numbers, by the MPN method after enrichment in Fraser broth according to the French standard AFNOR V08–055 [18]. MPN calculations were performed with the MPN Calculator software available at http://members.ync.net/mcuriale/mpn. Colonies with the typical morphology of Listeria were isolated and identified by the API Listeria biochemical assays (Biomérieux, Marcy-l'Etoile, France) and the hemolysin test according to the AFNOR V08-055 French regulation [18]. Genotype profiles were determined by M13-PCR to compare the identity of isolated strains with that of the strain used for inoculation. The PCR DNA template was prepared as described by Christiansson et al. [19] and M13-PCR was performed on 1 μl of DNA template as described by Guinebretière et al. [20]. To quantify spores of Clostridium, 5 ml of each sample was pasteurised at 80°C for 12 min, spread-plated on Differential Reduced Clostridial Medium (DRCM, Difco Laboratories, Detroit, USA) and incubated under anaerobic conditions at 30°C for 2–9 days. Colonies morphologically identical to those of the C. sporogenes inoculum were further isolated and identified with the API system (Anaerobe API ID 32A, Biomérieux).

2.4 Bacterial inoculum preparation

To prepare inoculum, one colony was inoculated in 10 ml of Tryptone Soya Broth (TSB, Oxoid, Basingstoke, UK) for L. innocua and anaerobic Tryptone Glucose Yeast broth (TGY) for C. sporogenes. TGY consisted of tryptone 30 g l⁻¹ (AES, Combourg, France), yeast extract 15 g l⁻¹ (AES), glucose 5 g l⁻¹, HCl-cystein 0.5 g l⁻¹ and resazurin 4 mg l⁻¹ adjusted to pH 7.3.

L. innocua was grown at 30°C under agitation (200 rpm) for 24 h. Bacterial cells were harvested by centrifugation (16,000g, 10 min), washed twice in sterile physiological water (9 g l⁻¹NaCl) and resuspended in the latter at the appropriate concentration. This inoculum was named “non-adapted inoculum”. An inoculum from L. innocua cells grown in compost-based medium, named “adapted inoculum”, was prepared in 2002 as follows. Compost-based medium consisted of a suspension of autoclaved compost (two successive heatings at 120°C for 1 h at 1-day intervals) in sterile physiological water at a ratio of 1/1 (w/w). This medium (2.1 kg) was inoculated with 10 ml of a suspension of washed cells of L. innocua in physiological water (10⁹ cells ml⁻¹) from a TSB-culture (30°C, 24 h) and incubated for 6 days at 20°C to produce the adapted inoculum.

The inoculum of C. sporogenes consisted of bacterial spores produced in two-phase medium according to the method of Anellis et al. [21]. After incubation at 30°C for 7 days, spores were harvested from the liquid phase by centrifugation, washed 5 times and stored in sterile physiological water at 4°C until used.

2.5 Inoculation of soil

In experimental fields, a total of 2 kg m⁻² of the amendments listed above (fresh weight before addition of inoculum), corresponding to 20 t ha⁻¹, was applied to each plot. For each inoculated plot, 5.5 kg of organic compost was mixed with 5.5 l of bacterial inoculum. When adapted Listeria inoculum was used, the compost (5.5 kg) was mixed with 700 g of adapted inoculum. For plots fertilized with non-inoculated compost, the bacterial inoculum was replaced by physiological water. After manual homogenisation in a plastic bag, inoculated and non-inoculated composts were manually poured onto the soil surface and incorporated into the soil by ploughing with a roto-till. Parsley was transplanted the following day.

Survival in soil of the strain of L. innocua (CIP 80–12), used in field experiment, was compared to that of the strains of L. monocytogenes (EGDe and LmP60) and L. innocua (LiP60) in the laboratory using soil microcosms with amendments. Soil from the experimental field, 97.2% dry weight, was dispensed into 100 g quantities in sterile Stomacher bags. Non-adapted inoculum (10 ml) was added to amendments (10 g) and mixed with 100 g of soil. The soil microcosms were incubated at 20 ± 1°C and analysed after 0, 3, 6 and 10 days as described above for field conditions. Three soil microcosms were inoculated for each strain.

2.6 Bacterial survival and transfer to plants under field conditions

Experimental plots of parsley were set up in the spring of 2000, 2001, 2002 and 2003 (3 April–21 June 2000, 10 April–3 July 2001, 28 March–25 June 2002, 31 March–2 June 2003) in fields of the INRA research center (Montfavet, Vaucluse, France). The soil contained 34% sand, 27% clay and 39% loam and had a pH of 8. Climatic conditions were representative of a typical French south-east climate with a minimal average temperature of 12 ± 1°C, a maximal average temperature of 24.5 ± 2°C and an average rainfall of 156 ± 43 mm during the crop period for all the years considered. Survival of Listeria was also measured during winter, with a minimum average temperature of 2.3 ± 3°C, a maximum average temperature of 10.7 ± 3°C and an average rainfall of 8 ± 2.3 mm.

Parsley seedlings at the 3-leaf stage were planted in field plots at a density of 8 plants/linear meter in rows spaced 40 cm apart. Each plot was surrounded by a 2 m wide buffer zone which did not receive any treatment. A wire fence surrounded the experimental field (135 m²) to keep out wild animals. Field plots were alternatively located on two neighbouring experimental fields each year. Three replicate plots of the three treatments tested each year were laid out according to a Latin square design. Treatments were: (1) control plots with no amendment, (2) plots fertilized with non-inoculated amendment (referred to as non-inoculated plots) and (3) plots fertilized with amendment inoculated with bacterial inocula (inoculated plots).

To evaluate bacterial survival in the soil, soil samples were collected between the parsley rows by scraping the soil surface (1–2 cm deep) with a sterile spatula. To detect and quantify bacterial populations on parsley, leaves were harvested using sterile scissors and tweezers. Four samples of three parsley leaves were randomly harvested in the three central rows of each plot. After each leaf sampling, all remaining old leaves were pruned from plants. This was done so that the bacterial populations detected on parsley would represent transfer of bacteria from the soil rather than spread of bacteria from leaves.

2.7 Evaluation of movement of bacteria from soil to plants via splashing, aerial dissemination and systemic movement in plant tissue

The presence of L. innocua in aerosol above an inoculated field plot was measured using an airborne particle sampler (High Throughput ‘Jet’ Spore sampler, Burkard Manufacturing Co Ltd, Rickmansworth, UK). Airborne bacteria were sampled after fertilization of a field plot (1 m²) with inoculated compost. Air samples of 8500 l were collected on Oxford Listeria selective agar medium in triplicate just after fertilization and daily for 5 days. The presence of Listeria colonies on Oxford plates was checked daily during incubation for 4 days at 30°C. This experiment was conducted twice.

The transfer of L. innocua via splashing was measured in a field plot (1.5 m × 1 m) fertilized with inoculated compost (10⁸ CFU g⁻¹) which was watered from a central point at 1.8 m height. Splashed particles were collected in Petri dishes (diameter 9 cm) containing 20 ml of sterile physiological water, placed around the impact point of water at 5 to 15 cm height. Water from Petri dishes and soil was analysed for Listeria populations by plating on selective medium.

Migration of L. innocua cells from the soil to the parsley leaves via the plant tissue was determined by cultivating parsley (220 plants) in heavily inoculated soil in a growth chamber. Commercial substrate (BF1, Tref, Vulaine, France) was inoculated with L. innocua at 3 × 10⁸ CFU g⁻¹ dry wt soil. To prevent contamination of the aerial parts of the parsley by splashing or by capillarity, the inoculated soil layer was covered with a 9-cm layer of non-inoculated soil and the soil surface was covered with plastic wrap. Then older leaves were cut and only emerging central leaves were left on the plants. Parsley was drip-irrigated daily to avoid splashing. Listeria was not detected in soil before plantation nor in the soil from the upper non-inoculated layer after plantation. All parsley leaves were harvested 11 days after transplanting for MPN enumeration of L. innocua.

2.8 Measurement of L. innocua survival on parsley leaf surfaces via direct inoculation

Three drops (3 × 15μl) of L. innocua non-adapted inoculum (10⁷ CFU g⁻¹) were deposited on individual parsley leaves on plants in the field. Four hundred leaves were inoculated in the early morning (7:30 AM) during sunny summer days. The experiment was conducted twice, in June 2002 (and July 2002) with an average minimum temperature of 14.2 ± 1.5°C (17.2 ± 1.6°C), an average maximum temperature of 29.3 ± 1.5°C (34.3 ± 2°C), an average minimum relative humidity of 27 ± 6.1% (23.6 ± 10.6%), an average maximum relative humidity of 97.3 ± 4.9% (96.9 ± 2.4%).

2.9 Statistical analysis

The fate of L. innocua and of C. sporogenes in experimental field was each tested in two independent experiments. For each experiment, results were expressed as the means ± standard error of the replicate samples (12 for plate counts, 45 for MPN counts). Standard errors were frequently below 0.1 log g⁻¹ and in such cases do not appear on the figures. The effect of factors was tested by ANOVA at the 95% level, using Systat software version 9 (SPSS Inc., Chicago, IL, USA).

3 Results

3.1 Survival of L. innocua and L. monocytogenes in soil and compost microcosms

All strains tested decreased over time in the soil microcosm at 20°C from approximately 10⁶ to 10⁴ CFU g⁻¹ after 10 days. The rate of decrease was calculated for each strain using linear regression (Table 1). L. innocua (CIP 80–12) behaved similarly on both kinds of compost and the two strains of L. innocua tested survived slightly better (rate of decline of −0.17 log d⁻¹) than the two strains of L. monocytogenes (rates of decline of −0.24 and −0.32 log d⁻¹). L. innocua appeared as a good surrogate to assess survival of L. monocytogenes in soil. No differences were observed among the two strains of L. innocua tested. Therefore, only strain CIP 80–12 of L. innocua was used in experimental fields.

1

The fate of two strains of Listeria innocua and of two strains of Listeria monocytogenes in soil microcosms with organic amendments

Listeria species	Strains	Rate of decrease (log d−1)	Initial inoculum (log CFU g−1)	Regression coefficient
Innocua	CIP 80–12	−0.168 (1)	5.8	0.98
Innocua	CIP 80–12	−0.172	6.5	0.96
Innocua	Li P60	−0.171	6.5	0.99
Monocytogenes	EGDe	−0.324	5.7	0.93
Monocytogenes	Lm P60	−0.243	6.7	0.99

Soil contained composted bovine manure, except ⁽¹⁾ which contained composted sewage sludge. Rates of decrease and regression coefficient were calculated by linear regression of log numbers of Listeria obtained on three replicate soil microcosms after 0, 3, 6 and 10 days at 20°C (P < 0.001 for all the linear regressions).

Survival in compost microcosms (normal, non-sterilized compost held at 20°C in the laboratory) of L. innocua (CIP 80–12) grown in a laboratory medium (non-adapted inoculum), was compared to that of the same strain grown in sterile compost (adapted inoculum). Survival of the adapted inoculum was greatly enhanced compared to that of the non-adapted inoculum: 20 days after inoculation, the population of the adapted inoculum lost only 0.5 log CFU whereas that of the non-adapted inoculum lost 1.8 log CFU. Because the history of the inoculum affected survival of L. innocua under laboratory conditions, both adapted and non-adapted inocula were tested under field conditions.

3.2 Behaviour of Listeria and Clostridium populations in the crop environment

The introduced populations of C. sporogenes declined only slightly (less than 0.7 log units), and were still detectable at a high level (4.7 log CFU g⁻¹ dry wt soil) 16 months after their introduction to the soil (Fig. 1), representing a rate of decline of −0.007 log d⁻¹. All the colonies counted as C. sporogenes from inoculated soil were confirmed to be C. sporogenes. Some sulfate-reducing spores were detected in non-inoculated and control soil, but none was confirmed as C. sporogenes. Spores of C. sporogenes were detected on parsley leaves grown in inoculated soil at concentrations ranging between 100 and 1000 CFU g⁻¹ through the entire cultivation period. Similar results were obtained in a duplicate experiment (not shown in Fig. 1).

1

Persistence in soil (closed squares and solid curve) and presence on parsley leaves (after each leaf sampling, all the remaining leaves were pruned) (dashes without line) of C. sporogenes spore populations in inoculated field plots. Open squares and broken curve corresponded to total bacterial counts in soil. C. sporogenes was not detected in the composts used to amend plots, nor in soil and parsley cultivated on control plots and plots amended with non-inoculated compost. Dashes below the horizontal dotted line represent numbers of C. sporogenes on parsley below the limit of detection (2.0 log CFU g⁻¹). Detection limit of C. sporogenes in soil was 4.2 CFU g⁻¹.

During the cultivation period, L. innocua populations in inoculated soil declined during the first 30 days from 4.0 × 10⁵ CFU to <10 CFU g⁻¹ dry weight soil, with a rate of decline of approximately −0.18 log d⁻¹ in 2002 (Fig. 2) and of approximately −0.17 log d⁻¹ in 2003 (results not presented). Interestingly, this rate of decline observed in the field was very similar to that observed in soil microcosms. Thereafter, the rate of decline decreased to approximately −0.04 log d⁻¹ in 2002 (−0.07 log d⁻¹ in 2003) and Listeria became undetectable (<0.25 MPN g⁻¹) after 90 days in 2002 (after 65 days in 2003). Listeria populations introduced to the field during winter survived better than those introduced in the spring with a decline of less than a factor of 1000 after 36 days, i.e., a rate of decline of approximately 0.08 log d⁻¹ (results not presented), compared to a decline of a factor 10⁵ in spring over the same period (Fig. 2). L. innocua was detected on parsley leaves until its population in soil dropped below 1 CFU g⁻¹, corresponding to the first 30 days of cultivation (Fig. 2). As determined by MPN enrichment enumeration, the level of contamination of leaves was very low, between 0.75 and 20 Listeria g⁻¹, and decreased over time to become undetectable (i.e., <0.1 MPN in 25 g) after 30 days. Molecular typing of Listeria colonies isolated from soil and parsley samples confirmed that these colonies were identical to the inoculated L. innocua strain. Fig. 2 presents results from 2002. In 2003, the rate of decline over the first 30 days was −0.17 log d⁻¹ and L. innocua was not detected on parsley leaves after 30 days in the experiment of 2003.

2

Persistence in soil (curves) and presence on parsley leaves (see Fig. 1) (symbols without curve) of L. innocua populations from a compost-adapted inoculum (dotted lines and ×) and from a non-adapted inoculum grown in rich broth medium (solid lines and dashes). Preparation of the inocula is described in the Material and Methods section. Dashes and × below the horizontal dotted line represent parsley leaves with number of L. innocua below the detection limit (0.04 MPN g⁻¹). Detection limit of L. innocua in the soil was 0.25 MPN g⁻¹. L. innocua was not detected in the composts used to amend plots, nor in soil nor on parsley cultivated in control plots nor plots amended with non-inoculated compost.

Survival in soil and transfer to parsley leaves of Listeria grown in sterile compost at 20°C before their introduction in the field environment (adapted inoculum) were compared to those obtained with the non-adapted Listeria inoculum (i.e., inoculum grown in rich medium at 30°C) used in previous experiments. However in the crop environment, the adapted L. innocua inoculum did not show overall better survival in soil nor greater transfer to leaves than the non-adapted inoculum (Fig. 2).

3.3 Evaluation of mechanisms of transfer of Listeria from soil to plants

Putative internalisation of L. innocua through the root system and migration to the leaves were tested on parsley transplants cultivated in highly contaminated soil (3.2 × 10⁶ CFU g⁻¹ dry wt soil). After 11 days of cultivation, the number of Listeria in the soil dropped to 3.1 × 10⁴ CFU g⁻¹ dry wt soil and Listeria was not detected by MPN enrichment in parsley leaf samples. The detection limit for these assays was 0.2 Listeria/25 g of parsley leaves. No colonies of Listeria were detected in air samples taken above field plots fertilized with inoculated compost, although the soil contained between 1.0 × 10³ and 3.6 × 10⁶ CFU g⁻¹ dry wt soil.

The number of Listeria that transferred via splashing from soil to aerial plant surfaces was estimated by simulating the impact of rainfall or sprinkling irrigation. L. innocua populations were detected in samples of splashed water. The numbers of L. innocua decreased significantly (P < 0.05) with increasing distance from the water impact point and with decreasing concentration of Listeria in the soil (Table 2). These results showed that splashing is a route of bacterial transfer from contaminated soil to parsley leaves under field conditions. It should be noted that either rain or sprinkler irrigation had occurred before each sampling of parsley leaves showed in Figs. 1 and 2.

2

Transfer of L. innocua populations by splashing from amended soil to aerial surfaces

Days after amendment with inoculated compost	Number of L. innocua in soil (log CFU g−1)	Observed numbers of L. innocua transferred (log CFU cm−2)a
		Values in bracket correspond to estimated number of L. innocua g−1 of parsley leavesb
		Distance from water impact point (cm)
35	40	50	65
1	6.8± 0.05	2.4 (4.4) ± 0.12	2.2 (4.2) ± 0.15	2.0 (4.0) ± 0.09	1.5 (3.6) ± 0.26
2	6.0± 0.28	1.8 (4.0) ± 0.15	1.7 (3.8) ± 0.09	1.6 (3.6) ± 0.08	0.8 (2.8) ± 0.08
4	5.1± 0.06	1.3 (3.5) ± 0.41c	1.7 (3.8) ± 0.23	1.6 (3.7) ± 0.15	1.3 (3.3) ± 0.13
5	3.1± 0.20	0.8 (2.9) ± 0.17d	1.3 (3.5) ± 0.41c	1.5 (3.9) ± 0.43c	1.2 (3.6) ± 0.72c

^aListeria were collected in Petri dishes containing 20 ml of physiological water and enumerated. Bold values represent the means ± standard error (n= 4, except for d= 35 cm n= 6).

^bPredicted numbers of Listeria g⁻¹ transferred on parsley leave were obtained by considering that a parsley leaf corresponds to three 2 cm-equilateral triangles and weighs 0.12 g.

^cOne negative sample for Listeria (≤0.5 log CFU cm⁻²). The value 0.5 log CFU cm⁻² was used for calculations for this sample.

^dThree negative samples for Listeria. See footnote c for calculations.

3.4 Survival of L. innocua on aerial parsley surfaces

L. innocua populations directly inoculated on parsley leaves decreased from 1.4 × 10⁷ to <0.23 CFU g⁻¹ within 48 h under field conditions. A rapid decrease of 6 log cycles occurred within the first 5 h, followed by a slower decline of the remaining fraction of the population (2 log cycles within 43 h). This minor fraction of the initial population was presumably more resistant to the parsley leaf surface conditions. However, in spite of the high inoculum applied to the leaves, no Listeria could be detected after 48 h.

4 Discussion

L. innocua and L. monocytogenes are naturally present in the environment and are frequently associated, suggesting that these two species have similar ecological requirements [3,22,23]. Here, we assume that the behaviour of the surrogate L. innocua is similar to that of the pathogenic species L. monocytogenes. This is supported by our results in soil microcosms suggesting that the ability of L. innocua to survive in soil is similar or slightly higher than that of L. monocytogenes. Hence, our results suggest that populations of L. monocytogenes introduced in the soil through composts could transfer to crops.

In spite of the high inoculum doses used in this study, L. innocua was not able to persist in the soil for more than 90 days and was not detected on parsley leaves beyond 30 days after planting. The use of laboratory-maintained strains of L. innocua may explain the poor persistence of these species observed under environmental conditions. However, in our study, the pre-adaptation of L. innocua inoculum to compost did not increase its survival in the field environment. In soil microcosms we showed that strains of L. innocua and L. monocytogenes recently isolated from organic amendments did not survive better than the L. innocua strain from a culture collection. In previous works, L. monocytogenes also disappeared rapidly from soil amended with pig manure [5]. Listeria was detected in soil 90 days after inoculation. This is consistent with previous studies showing that L. monocytogenes persisted in soil for up to 8 weeks after amendment with naturally contaminated sewage sludge [16].

Several surveys detected L. monocytogenes in soil and vegetation [1–3]. L. monocytogenes was found in only one sample out of 136 samples of urban soil in England [3], indicating that L. monocytogenes presumably does not survive for long periods in soil. Welshimer and Donker-Voet [1] isolated L. monocytogenes from soil and vegetation in April, particularly on decaying vegetation, but never in September. The authors concluded that L. monocytogenes was not able to survive in soil during spring and summer. Similarly, we observed that L. innocua survived better in winter than in spring. In contrast, Weis and Seeliger [2] isolated L. monocytogenes from many samples of soil and vegetation, taken from both cultivated and uncultivated areas. Uncultivated fields (44% of samples found positive) and mud (31.5% samples found positive) were particularly contaminated. However, frequent incidence of L. monocytogenes could be the result of either long survival or repeated contamination from an external source. The authors observed survival of the same serovar in the same place for half a year only in faded and decaying grass. In our experimental fields, decaying vegetation and/or mud were not present, which might explain the poor survival of Listeria.

Similar rates of decline of Listeria in soil over the first 30 days were observed in the experimental fields in spring and in the laboratory microcosms. Because environmental conditions were very different between laboratory and field, and because two field experiments were done in two different years with different climatic conditions, it can be assumed that the rate of decline is mostly due to some intrinsic factor of the soil. Soil acidity and texture influence survival of Listeria spp. In the laboratory, L. monocytogenes survived longer in loamy soil or alkaline soil than in sandy or acid soil [11–14]. It also survived better in agricultural soil than in clay soil [13]. In our study the soil was a rich agricultural soil containing loam and was alkaline, all conditions favourable for Listeria spp. survival. Several studies showed that L. monocytogenes was able to grow in autoclaved soil but decreased in non-autoclaved soil [11,12,24]. It can therefore be assumed that the soil microbiota is the main cause for the decline of Listeria spp. in our work. Similarly, we found that L. innocua declined in normal compost but multiplied in the sterilised compost used to produce inoculum.

L. innocua survived better in winter than in spring. In laboratory, Welshimer [13] showed that allowing soil to dry reduced Listeria survival. In this study, Listeria survived similarly in soil microcosms kept in sealed plastic bags and under field conditions in spring, where humidity of the soil surface was variable, from dry periods to saturation after rainfall. In addition, winter was dry (only 8 mm rainfall) as compared to spring (156 mm rainfall). Soil humidity was presumably not a key factor for survival of Listeria under our conditions. In contrast, temperatures in winter were below 10°C, whereas in spring, during parsley production, temperature was frequently above 20°C. Temperature presumably had an important role on Listeria survival in soil. This is consistent with previous work done in soil microcosms which showed better survival of L. monocytogenes in soil microcosms at low (≤5°C) than at high (15–21°C) temperatures [14,24].

L. innocua was not detected in internal tissues of parsley, in contrast to observations of Escherichia coli in lettuce seedling [25] and Salmonella in tomato plants [26,27]. This could be due either to a general inability of L. innocua to enter vascular tissues of plants or to a specific inhibitory effect of parsley. The furanocoumarins psoralen, 8-methoxypsoralen and 5-methoxysporalen have been found in parsley leaves from various cultivars, in concentrations largely exceeding their minimum inhibitory concentration on L. monocytogenes and L. innocua[28]. Hashem and Sahab [29] attributed antibacterial activity of parsley on Gram-positive bacteria to the coumarin pereflorine B. Steam distillates from parsley leaves were among the most inhibitory against L. monocytogenes, out of a large range of herbs and spices tested [30]. Parsley clearly has antimicrobial activity against Listeria spp. However, all these studies were performed in vitro, using compounds chemically extracted from parsley leaves. There was no indication that these compounds would come into contact with Listeria spp. in undamaged parsley plants. In vivo, furanocoumarins from parsley are known phytoalexins, i.e., involved in the defence mechanisms against plant pathogenic microorganisms [31]. However, there was no indication that L. monocytogenes could trigger parsley defence mechanisms and could be exposed to the furanocoumarin phytoalexins.

In this study, Listeria was no longer detected on parsley leaves a month before it disappeared from soil. For fresh produce consisting of aerial parts of plant, this may represent an additional safety factor against contamination with L. monocytogenes from soil amendments. It is important to note that the limit of detection on parsley leaves (0.04 MPN g⁻¹) was 10 times lower than the lowest acceptable limit for L. monocytogenes in foods in the EU (1 in 25 g). The major route of transfer of Listeria from soil to aerial parts of plant involved splashing. This suggests that the use of protective mulches such as plastic films and the use of drip irrigation to prevent microbial splashing from soil to produce could further limit the risk of contamination of crops. However the effect of mulches or plastic films on L. monocytogenes is not known.

The levels of spores of C. sporogenes remained relatively constant during the course of the experiment, suggesting that they survived rather than proliferated in soil and on parsley. Nonetheless, their long-term persistence (>1 year) in amended soil and their transfer to parsley leaves throughout the cultivation period (8–10 weeks) constitute an important risk of crop contamination.

In European countries, the use of amendments based on animal manure and sewage sludge for crop fertilization is increasingly monitored by Regulations and/or Codes of Practice that are increasingly restrictive with regard to microbiological agents [32,33]. This study involved thorough quantitative assessment of Listeria survival in a crop environment, using very low detection limits. The results suggest that delays imposed between application of organic fertilizers and cultivation should prevent contamination of produce with L. monocytogenes from amendments. As an example, a delay of 10–18 months is currently required by the French Regulations [33]. This level of protection would be amply sufficient with regard to L. monocytogenes under the soil and climate conditions employed in this study. Conversely, the presence of spores of C. botulinum in the crop environment would probably not be prevented by such delays. The increasing use of composted wastes as crop fertilizers and the absence of regulation limits concerning pathogenic spore-forming bacteria in these fertilizers may enhance the risk of produce contamination by these pathogens.

Acknowledgements

This work received support from the French Ministry of the Environment under a contract “Environnement – Santé” and from the French Ministry of Agriculture under a contract “Aliment – Qualité– Sécurité”. Part of this work was performed by Nicolas Dreux in the course of his PhD, for which he received fellowships from the “Institut National de la Recherche Agronomique” and from the “Conseil Régional Provence Alpes Côte d'Azur”. We are grateful to Dr. Hélène Bierne, Institut Pasteur Paris, France for providing L. monocytogenes strain EGDe.