Effects of Cover Crop Species and Season on Population Dynamics of Escherichia coli and Listeria innocua in Soil

Abstract

Cover crops provide several ecosystem services, but their impact on enteric bacterial survival remains unexplored. The influence of cover cropping on foodborne pathogen indicator bacteria was assessed in five cover crop/green manure systems: cereal rye, hairy vetch, crimson clover, hairy vetch-rye and crimson clover-rye mixtures, and bare ground. Cover crop plots were inoculated with Escherichia coli and Listeria innocua in the fall of 2013 and 2014 and tilled into the soil in the spring to form green manure. Soil samples were collected and the bacteria enumerated. Time was a factor for all bacterial populations studied in all fields (P < 0.001). E. coli levels declined when soil temperatures dipped to <5°C and were detected only sporadically the following spring. L. innocua diminished somewhat but persisted, independently of season. In an organic field, the cover crop was a factor for E. coli in year 1 (P = 0.004) and for L. innocua in year 2 (P = 0.011). In year 1, E. coli levels were highest in the rye and hairy vetch-rye plots. In year 2, L. innocua levels were higher in hairy vetch-rye (P = 0.01) and hairy vetch (P = 0.03) plots than in the rye plot. Bacterial populations grew (P < 0.05) or remained the same 4 weeks after green manure incorporation, although initial reductions in L. innocua numbers were observed after tilling (P < 0.05). Green manure type was a factor only for L. innocua abundance in a transitional field (P < 0.05). Overall, the impacts of cover crops/green manures on bacterial population dynamics in soil varied, being influenced by bacterial species, time from inoculation, soil temperature, rainfall, and tillage; this reveals the need for long-term studies.

INTRODUCTION

In agricultural environments, soil can serve as a reservoir and route of transmission for foodborne pathogens (1). Since fresh produce is often consumed raw without a “kill step” to inactivate human-pathogenic microorganisms, it is important to prevent the contamination of these foods during production (2). Growers may implement good agricultural practices (GAPs) to minimize the risk of produce contamination at the preharvest level (3), as this program focuses on on-farm risk factors, including animal-based fertilizers, irrigation water quality, farm worker training and hygiene, and wildlife exclusion. Outbreaks and contamination events, however, continue to occur, emphasizing the need to evaluate the role of other agricultural practices, which thus far have received less attention, on enteric pathogen dynamics on a farm.

One such practice is cover cropping, the establishment of a crop, typically a small grain or legume, in between cultivations of a cash crop. In recent years, economic and environmental considerations have renewed interest in this old practice for improving crop productivity and soil health and maintaining the sustainability of agroecosystems (4). Cover cropping brings a variety of ecosystem services to agricultural systems, including seasonal protection of soil from erosion, weed suppression, soil improvement, and nutrient management though nitrogen fixation and carbon accumulation (5). From a microbial perspective, the establishment of a cover crop provides a rhizosphere effect (6), whereby plant roots modify the soil habitat as they grow, improving aeration and serving as a source of nutrients to microorganisms, leading to enhanced microbial growth and activity (7). Consequently, cover cropping has a positive effect on soil microbial populations, processes, and activities (8). In comparison to the biomass and activities of microorganisms in bulk soil, those of soilborne microorganisms are boosted as a result of the exudation of phytocompounds by plant roots (9). In long-term field trials in which different agricultural systems were compared, soils from organically managed plots had increased microbial biomass and biodiversity, enhanced soil fertility, reduced soil erosion, and improved soil quality, cycling efficiency, and overall crop productivity (10, 11). Whether there is a positive rhizosphere effect from a variety of cover crop species on the growth, persistence, and activity of foodborne pathogens in soil, however, has not been evaluated.

Green manure, which is formed by the incorporation of a fall-planted cover crop into the soil in the spring, is employed to increase soil organic matter (reviewed in reference 12) and soil quality (13), stimulate soil microbial growth and activity, which enhances the subsequent mineralization of plant nutrients (14), and increase soil fertility (15). Cover crops and green manures have also been used to suppress soilborne and foliar plant diseases (16), plant-parasitic nematodes (17), and insect pests (reviewed in reference 18). For example, a hairy vetch (HV) cover crop (Vicia villosa Roth) suppressed multiple plant diseases on watermelon (19, 20) and pumpkin (21) and reduced foliar necrosis in processing tomato (22). Ryegrass (Lolium multiflorum Lam) green manures reduced powdery scab in potatoes (23) and Verticillium microsclerotia density in soil in cauliflower fields (24). While there are many instances in which cover crop-suppressive effects on plant pathogens have been documented, their potential biocontrol activity on enteric foodborne pathogens has not been explored.

The objective of this study was to evaluate the impact of single-species and multispecies cover crop mixtures on bacteria relevant to food safety. The influence of cover cropping on foodborne pathogen indicators was assessed in five cover crop/green manure systems: cereal rye (R), hairy vetch, crimson clover (C), a hairy vetch-rye mixture (HVR), a crimson clover-rye mixture (CR), and a no-cover crop bare-ground (BG) control. The study was conducted over a 2-year period in an organic field and over a 1-year period in a conventional field that was transitioning to organic production.

MATERIALS AND METHODS

Field preparation.

Field experiments were conducted on certified organic and transitional (previously conventional) land at the University of Maryland (UMD) Lower Eastern Shore Research and Education Center (LESREC) in Salisbury, MD (about 38°N and 75°W). Both fields had low organic matter (<1%) and dimensions of 107 m long by 27 m wide (∼0.3 ha in size). The organic field was composed of Fort Mott and Rosedale loamy sand soils (0 to 5% slope and pH 6.8), was within 40 m of a woodland conservation buffer, and had a history of organic mixed-vegetable production. The transitional field was composed of Fort Mott loamy sand soils (0 to 2% slope and pH 5.9), surrounded by other fields, and had a history of conventional agronomic crop (primarily field corn and soybean) production. The trial was conducted three times: in 2013 to 2014 and 2014 to 2015 in the organic field and in 2014 to 2015 in the transitional field. Cover crops were sown using a grain drill on 22 October 2013 (year 1) and 23 September 2014 (year 2), and overhead irrigation was applied briefly to improve seed germination and establishment. Bare-ground (BG) plots served as the control and remained fallow throughout the fall, winter, and spring seasons. The field experiments were arranged in a randomized complete block design with four replications in year 1 (organic field only), with an additional replication (n = 5) in year 2 (organic and transitional fields). The individual plots were 27 m long by 3 m wide. Deer fencing was installed after cover crop seeding to deter wildlife intrusion. At the flowering stage, cover crops were tilled into the soil to a depth of 15 to 20 cm using a moldboard-type plow (15 May 2014 for year 1 and 6 May 2015 for year 2). The fields were plowed again on 3 June 2014 (year 1) and 22 May 2015 (year 2) to further incorporate the green manure into the soil.

Cover crops.

Cover crop treatments included hairy vetch (V. villosa Roth) (HV) at 18.14 kg/ha, crimson clover (Trifolium incarnatum) (C) at 9.07 kg/ha, cereal rye (Secale cereale L.) (R) at 31.75 kg/ha, a 2:3 (wt/wt) mixture of hairy vetch (9.07 kg/ha) and rye (13.60 kg/ha) (HVR), and a 2:1 (wt/wt) mixture of crimson clover (9.07 kg/ha) and rye (4.54 kg/ha) (CR). The C and CR plots were planted in the second year only. Certified organic (when available) or non-genetically modified organism (GMO)-treated cover crop seed was purchased from the following suppliers: Johnny's Selected Seeds (Winslow, ME), High Mowing Organic Seeds (Wolcott, VT), Fedco Seeds (Waterville, ME), and Territorial Seed Company (Cottage Grove, OR). Legume cover crop seed (hairy vetch and crimson clover) was mixed with Organic Materials Review Institute (OMRI)-listed N-Dure inoculant (INTX Microbials, LLC, Cary, NC) prior to being seeded to encourage nitrogen-fixing nodule formation.

Bacterial indicators.

Nonpathogenic strains were used for introduction into field plots as bacterial indices for enteric pathogens. Generic Escherichia coli strains isolated from liquid dairy manure collected from a dairy farm in Clarksville, MD, and Listeria innocua (ATCC 33090) were used. E. coli was incubated in Trypticase soy broth (TSB) (BD Diagnostic Systems, Franklin Lakes, NJ) for 24 h at 37°C. L. innocua was incubated in brain heart infusion (BHI) broth (BD) for 48 h at 30°C. The cells were centrifuged for 10 min at 2,000 × g, the supernatant was discarded, and pellets were resuspended and washed with phosphate-buffered saline (PBS). After the pellets were washed, 10 ml of 0.1% peptone water (PW) was added to the pellets and the suspension was vortexed. The concentrations of the stock suspensions were confirmed by serial dilutions in 0.1% PW. The dilutions were plated onto Trypticase soy agar (TSA) (BD) plates for generic E. coli and brain heart infusion agar (BHIA) (BD) for L. innocua. A cocktail of the two bacteria (∼10⁶ CFU/ml) was prepared for field inoculations. The inoculum was kept on ice until application to avoid any growth while in transit.

Field inoculation.

The bacterial cocktail was applied to the field 6 days after cover crop seeding in year 1 and 8 days after cover crop seeding in year 2. In year 1, the bacterial suspension was adjusted to a final concentration of ∼10⁶ CFU/ml; this was sprayed evenly on the soil surface of the treatment plots using a handheld sprayer boom with four nozzles, with a horizontal spray width of 1.8 m and application at a rate of ∼8 ml/m². In year 2, the applied concentration was adjusted to account for levels already present in the soil. The applied concentrations were 10⁴ CFU/ml E. coli and 10³ CFU/ml L. innocua.

Field sample collection.

Soil samples were collected immediately prior to field inoculation and within 2 h following inoculation. Sampling continued every 2 weeks postinoculation until frost, monthly thereafter until cover crop incorporation into the soil, and then biweekly for 4 weeks. Field soil samples (∼100 g of composite soil) consisted of four soil subsamples per cover crop treatment per replicate per field and were collected in sterile Whirl-Pak bags (Nasco, Ft. Atkinson, WI) using 60-ml sterile scoops (Fisher Scientific, Hampton, NH). In the cover crop treatments, samples were collected to a 7-cm depth from the root zone and were nondestructive to the cover crops. In the bare-ground treatments, soil samples were collected in areas devoid of weeds to a 7-cm depth. Latex gloves and plastic boot covers were worn for sample collection, changed between fields, and disinfected with 70% ethanol between samples. The samples were sealed, transported in coolers with ice packs, and delivered the same day or shipped overnight to the laboratory for microbiological analysis within 24 h. A total of 224 field soil samples were collected in year 1 (56 composite samples for each of the four cover crop treatments). In year 2, a total of 780 soil samples were collected (65 composite samples for each of the six cover crop treatments in each of the two field locations).

Sample preparation and bacterial enumeration.

Thirty grams of soil was mixed with buffered peptone water (BPW) (1:5 [wt/vol]), stomached at 200 rpm for 1 min, and allowed to recover for 1 h to revive injured cells. Two 96-well plates for a 3-tube most-probable-number (MPN) analysis were prepared, one for E. coli with brilliant green bile broth with 4-methylumbelliferyl-β-d-glucuronide (BRILA-MUG) (Criterion, Santa Maria CA) and one for Listeria using buffered Listeria enrichment broth (BLEB) (BD). The samples were shaken by hand for 10 s to homogenize them, serial dilutions were then prepared in the respective medium, and E. coli plates were incubated at 42°C for 24 h and Listeria plates at 30°C for 48 h. Following incubation, 10 μl of culture was either plated on Tryptone bile glucuronic agar (TBX) (Hi Media, Mumbai, India) and incubated at 42°C for 24 h for E. coli or plated on Oxford Listeria agar (OXA) (EMD Millipore, Billerica, MA) and incubated at 30°C for 24 to 48 h for L. innocua. The MPN protocol was modified from that used by Ingram et al. (25).

Meteorological data.

On-site weather data were recorded every 60 min using a CR1000 data logger (Campbell Scientific, Logan, UT) equipped with the following instruments: reflectometer (model CS650), anemometer (model 034B), pyranometer (model LI-COR), rain gauge (model TE525WS), and temperature and humidity probes (models 107 and HMP60, respectively). Soil temperature was evaluated on the day of sampling and recorded as the average of the maximum and minimum temperature readings.

Statistical analysis.

The MPN was calculated using the MPN Calculator, version VB6 (http://www.i2workout.com/mcuriale/mpn/). The data were log transformed and analyzed using a mixed model with repeated measures. Fixed effects were cover crop species and day postinoculation, while subject was treated as a random effect, due to repeated measures and nonindependence of the observations and residuals. Least-squares means were calculated, along with the standard error. Statistically significant effects at a P value of <0.05 were further analyzed with Tukey's honestly significant difference (HSD) test. All analyses were performed using JMP Pro version 11 (SAS Institute, Inc., Cary, NC).

RESULTS

Bacterial population dynamics in organic and transitional fields.

In the first year of the study from October 2013 to May 2014, cover crops were grown using organic practices, and soil samples were collected over a 27-week time period, at which time cover crops were incorporated into the soil. In year 2, cover crops were grown in two fields (one organic and one transitional) from October 2014 to June 2015, and soil sampling occurred over a 30-week period before the tilling. In both years, sampling of the green manures continued for an additional 4 weeks. Prior to field inoculation with the bacterial cocktail in the fall of year 1, E. coli was found to be mostly undetectable in the soil in the organic field but sporadically present at a mean of 0.4 log MPN/g of soil in two plots. L. innocua was undetectable in all plots. Following inoculation, the mean levels of bacteria in the organic field were approximately 6 log MPN/g of soil in year 1 and 4 log MPN/g of soil in year 2 (Fig. 1 and 2 and Table 1). In the transitional field in year 2, the mean bacterial levels following inoculation were 3 log MPN/g of soil for E. coli and 7 log MPN/g of soil for L. innocua (Fig. 1C and 2C and Table 1). In all cases, bacterial levels declined from fall to late spring (Table 1). E. coli levels declined to below the detection limit by weeks 5 and 9 in the organic field in the two consecutive years, respectively (Fig. 1A and B), and by week 5 in the transitional field (Fig. 1C). Although L. innocua levels diminished overall, this species persisted throughout the study period, dipping below the detection limit only in week 23 in the organic field in year 2 (Fig. 2B) and in week 30 in the transitional field (Fig. 2C).

FIG 1.

Population dynamics of E. coli in the organic field (A and B) and the transitional field (C) as impacted by cover crop, namely, hairy vetch (HV), crimson clover (C), rye (R), hairy vetch-rye (HVR), crimson clover-rye (CR), and a bare-ground no-cover-crop control (BG) in years 1 (A) and 2 (B and C) of the study. The data are expressed as the mean log MPN/g of soil from independent replicates. LOD, limit of detection.

FIG 2.

Population dynamics of L. innocua in the organic field (A and B) and the transitional field (C) as impacted by cover crop, namely, hairy vetch (HV), crimson clover (C), rye (R), hairy vetch-rye (HVR), crimson clover-rye (CR), and a bare-ground (BG) no-cover-crop control in years 1 (A) and 2 (B and C) of the study. The data are expressed as the mean log MPN/g of soil from independent replicates. LOD, limit of detection.

TABLE 1.

Mean differences in E. coli and L. innocua populations in organic and transitional fields between the time of inoculation and the time just prior to tillage^a

Yr and field	Wk	Log MPN/g of soil (mean ± SE)a
		E. coli	L. innocua
2014
Organic	0	6.15 ± 0.21 A	6.47 ± 0.23 A
	27	0.57 ± 0.2 B	4.61 ± 0.23 B
2015
Organic	0	3.62 ± 0.22 A	4.79 ± 0.28 A
	30	0.000 B	1.82 ± 0.29 B
Transitional	0	3.20 ± 0.21 A	7.55 ± 0.24 A
	30	0.000 B	0.22 ± 0.24 B

^aDifferent letters between the first and last week denote a statistically significant difference (P < 0.001) for that field.

Effect of cover crop species on bacterial population dynamics in an organic field. (i) E. coli.

In the organic field in year 1, both cover crop treatment (P = 0.004) and time (P < 0.001) were significant factors for E. coli survival. A statistically significant interaction between cover crop and time (P < 0.001) denoted that the effect of cover crop was most significant in the first 3 weeks of the study (Fig. 1A). E. coli populations in the bare-ground and HV plots displayed sharp declines in the survival of E. coli after week 1, while E. coli levels in R and HVR plots persisted at around 4 to 5 log MPN/g of soil until week 3 before declining. HVR and R plots supported the highest mean E. coli populations at 1.85 ± 0.15 and 1.82 ± 0.15 log MPN/g of soil, respectively, and were both significantly different (P = 0.007 and <0.01, respectively) from the lowest levels detected in bare-ground plots at 1.01 ± 0.15 log MPN/g of soil. In year 2, time (P < 0.001) was a significant factor in E. coli survival (Fig. 1B and C), while no significant differences among cover crops were detected. An interaction (P < 0.001) between the two factors revealed that the C, CR, and HV plots had higher E. coli levels in the fall than in other seasons, with overall means of 1.31 ± 0.23, 1.14 ± 0.23, and 1.10 ± 0.23 log MPN/g of soil. The lowest levels were recorded from the R plots (mean of 0.70 log MPN/g of soil).

(ii) L. innocua.

Time was also a factor for L. innocua persistence (P < 0.001) in year 1, while cover crop was not (Fig. 2). However, an interaction between cover crop and time (P < 0.001) revealed increased L. innocua persistence in the first few weeks of the study in the R and HVR plots (Fig. 2A). The HV plots exhibited the sharpest declines. The highest mean L. innocua levels were recovered from the R plots (4.56 ± 0.20 log MPN/g of soil), and the lowest were recovered from the HV plots (3.85 ± 0.20 log MPN/g of soil). In year 2, both cover crop (P = 0.011) and time (P < 0.001) had a significant impact on L. innocua survival (Fig. 2A). A significant interaction (P = 0.013) highlighted the dependency of the two factors. The levels of L. innocua cells were significantly higher in HVR (P = 0.010) and HV (P = 0.029) plots than in R plots (Fig. 2B). The mean levels were determined to be 3.04, 2.91, and 1.93 ± 0.21 log MPN/g of soil, respectively.

Effect of field management on bacterial population dynamics.

To determine whether field management (i.e., transition from conventional to organic production) would influence indicator bacterial population dynamics, a field previously under conventional management was included in the study in year 2. Prior to cover crop seeding in the fall of 2014, this field had been in a 3-year agronomic crop rotation (corn followed by wheat in 2011, double-crop soybeans in 2012, and full-season soybeans in 2013). Time (P < 0.001), but not cover crop, was a factor for E. coli and L. innocua survival (Fig. 1C and 2C). A significant interaction of time and cover crop (P < 0.001) indicated higher E. coli levels in the HVR and CR plots at certain time points. The HVR plots supported the highest mean population, at 1.05 ± 0.21 log MPN/g of soil, while the R plots had the lowest, at 0.39 ± 0.21 log MPN/g of soil. Survival of L. innocua was higher in bare-ground plots (4.11 ± 0.22 mean log MPN/g of soil) than in other cover crop treatments. The lowest population was detected in the R plots (3.20 ± 0.22 mean log MPN/g of soil). Differences were not statistically significant.

Effect of tillage and green manure on bacterial population dynamics. (i) E. coli.

Following week 27 in year 1 and week 31 in year two, the cover crops were tilled into the soil, resulting in a green manure. A comparison of bare-ground and cover crop or green manure soil in the organic field before and after tillage revealed no significant difference in the persistence of E. coli that could be attributed to cover crop treatment. In the transitional field, a very weak effect was detected, with the highest populations recorded in CR, HVR, and HV green manures, in decreasing order (P < 0.071) (Table 2). In the organic field during year 2, a statistically significant increase in E. coli populations was observed following tillage (P < 0.001) (Table 2).

TABLE 2.

Impact of tilling in green manure on bacterial population levels in soil

Organism	Yr	Field	Log MPN/g of soil (mean ± SE)a			P value
			Before tillage	2 wk after tillage	4 wk after tillage
E. coli	1	Organic	0.57 ± 0.30	0.23 ± 0.30	0.59 ± 0.30	0.699
	2	Organic	0.00 ± 0.21 A	0.29 ± 0.21 A	1.23 ± 0.21 B	0.001
	2	Transitionalb	0.00 ± 0.13	0.20 ± 0.13	0.42 ± 0.13	0.117
L. innocua	1	Organic	4.61 ± 0.15 A	3.57 ± 0.15 B	4.14 ± 0.15 A	0.001
	2	Organic	1.82 ± 0.29 A	0.73 ± 0.29 B	1.82 ± 0.29 A	0.016
	2	Transitionalc	0.22 ± 0.23 A	0.62 ± 0.23 A	1.97 ± 0.23 B	<0.001

^aDifferent letters denote statistically significant differences among time points.

^bThe cover crop treatment was a significant factor (P < 0.1).

^cThe cover crop treatment was a significant factor (P < 0.05).

(ii) L. innocua.

In the organic field, tillage negatively impacted L. innocua populations. Declines of about 1 log MPN/g of soil were recorded in both years, with a subsequent recovery to pretillage levels by week 4 posttillage (P < 0.05) (Table 2). In the transitional field, posttillage levels were significantly higher than pretillage levels (P < 0.001). Cover crop treatment was a significant factor only in the transitional field (P = 0.027) (Table 2), with mean R populations being 1.71 log MPN/g of soil higher than HVR populations (P = 0.011).

Effect of soil temperature on bacterial persistence. (i) E. coli.

In general, declines in the E. coli population appeared to coincide with decreasing soil temperatures as the fall season progressed, with resurgence observed in the spring when soil temperatures climbed above freezing. Following inoculation in year 1, E. coli populations persisted at >3 log MPN/g of soil in all cover crop plots at soil temperatures of >10°C (Fig. 3A). However, in bare-ground plots, a decline of >4.5 log MPN/g of soil by week 2 was observed despite temperatures being ≥10°C. As temperatures fell to <5°C, a 5- to 6-log MPN/g of soil decrease was observed after week 3, even in the cover crop plots. When temperatures warmed up again in the spring of the following year, E. coli resurged, but at low levels, never exceeding 1.72-log MPN/g of soil. In year 2, a similar decline in E. coli populations was seen in both bare-ground and all cover crop plots. E. coli levels hovered between 1.5 and 2 mean log MPN/g of soil at temperatures around 10°C during weeks 5 and 7 (Fig. 3B). The populations declined further as temperatures fell to <10°C. E. coli levels increased when temperatures were >20°C in the bare-ground and cover crop plots in week 34. Although the E. coli population in the transitional field exhibited a trend similar to that of the organic field, no significant increase in bacterial populations was observed in the spring in either the bare-ground or cover crop plots.

FIG 3.

Soil temperatures, three previous days' rainfall sums, and population levels of E. coli (A and B) and L. innocua (C and D) in soil from bare-ground (BG) or cover crop plots in year 1 (A and C) and year 2 (B and D) in an organic field. The population data are expressed as the mean log MPN/g of soil. The cover crop data are the combined log MPN values for all treatments.

(ii) L. innocua.

L. innocua persistence was independent of soil temperature, and bacterial populations continued to flourish throughout the study, regardless of season (Fig. 3D to F). At around 10°C, L. innocua levels persisted at around 4 to 5 log MPN/g of soil. In year 1, bacterial populations were highest when temperatures fell to 5°C in week 11 (Fig. 3D), when mean increases of 3.03 and a 4.74 log MPN/g of soil were observed in bare-ground and cover crop soil, respectively. Bacteria persisted at 1.4 to 4 mean log MPN/g of soil in year 2 at the lowest soil temperatures (Fig. 3E and F). Warming temperatures did not coincide with the increases in L. innocua. Populations either remained constant (Fig. 3D) or declined (Fig. 3E and F) with increasing temperatures in spring.

Effect of total precipitation on bacterial persistence.

The total precipitation for the 3 days prior to each sampling was summated to assess the effect rainfall had on bacterial population dynamics. In year 1, E. coli levels were higher in the absence of precipitation early in the sampling period (weeks 1 to 3), but declines coincided with a temperature dip rather than rainfall (week 5) (Fig. 3A). Population declines coincided directly with precipitation only in year 2 in sampling week 1 (4.8 mm of rain) and week 23 (9.4 mm of rain) (Fig. 3B). A resurgence of E. coli was observed only after the week 1 rain event. Similarly, L. innocua populations persisted regardless of precipitation in year 1, despite a 26.7-mm depth of rain recorded in week 7 and 36.6 mm in week 29 (Fig. 3C). In contrast, declines in the populations coincided with specific rain events in year 2 in weeks 1 and 23 but not week 30 (Fig. 3D). Population levels increased following the week 1 and week 23 events. In the transitional field, lower levels of E. coli and L. innocua were also recorded following the rain events of sampling weeks 1, 23, and 30, followed by bacterial population increases (Fig. 4).

FIG 4.

Soil temperatures, three previous days' rainfall sums, and population levels of E. coli (Ec) and L. innocua (Li) in soil from bare-ground (BG) or cover crop plots in year 2 in a transitional field. The population data are expressed as the mean log MPN/g of soil. The cover crop data are the combined log MPN values for all treatments.

DISCUSSION

This study examined the influence of cover crops on the dynamics of foodborne pathogen indicator bacteria in soil in a vegetable-producing area. The results from this study reveal that cover crops may influence E. coli and L. innocua population dynamics. The hairy vetch-rye cover crop mixture tended to support the highest bacterial populations, although the levels were sensitive to sampling date, and increases were not always statistically supported. The other multispecies cover crop treatment of crimson clover and rye also appeared to have a weak stimulatory effect on E. coli but not L. innocua. Monoculture cover crop treatments of hairy vetch, a legume that boosts nitrogen levels in soil (reviewed in reference 4), gave varied results but appeared to have an impact on bacterial population levels equivalent to those of the bare-ground plots in the first year. Rye monocultures, which are known to accrue soil carbon over time (26, 27), supported larger populations of both E. coli and L. innocua in year 1 while being the cover crop treatment harboring the lowest bacterial levels in year 2. Biomass variability has been reported for monocultures of rye compared to mixtures with hairy vetch (28). This might account for the variability in ecosystem services linked to nutrient availability, which in turn impacts bacterial population dynamics. Likely, the impact of the cover crop treatment on the indicator bacterial populations was further complicated by the interaction between the cover crop plant species, local environment, and timing of management practices (reviewed in reference 29).

Although one of the objectives of this study was to explore the biocontrol potential of cover crops against enteric bacteria, the hairy vetch-rye mixture favored the persistence of bacterial populations. Recalcitrance of enteric pathogens as a result of cover cropping is undesirable from a food safety perspective, although maybe not surprising, since a robust rhizosphere is expected to exert a strong rhizosphere effect. Understanding the physicochemical niche established in various cover crop rhizospheres will be critical for devising effective cover crop application and management practices that augment their biocontrol potential or attenuate the promotion of undesirable plant and human pathogens. Additionally, although this study did not consider rotations, the long-term benefits of certain cover crop species, such as carbon accumulation from rye, might also contribute to bacterial population fluxes. The favorable environment offered by the rhizosphere is expected to preferentially benefit microorganisms that are better adapted to soil than most enteric bacteria. This would explain the rapid declines seen for E. coli, as opposed to the more ubiquitous and persistent L. innocua levels. Ultimately, an understanding of the functions and interactions of various microorganisms in cover crop rhizospheres, be they antagonistic, competitively exclusive, or mutualistic, is needed for the optimization of cover crops for biocontrol purposes, although such data for enteric pathogens do not currently exist.

Cover crop treatment effects were inconsistent over the 2 years, exhibiting interactions with time and suggesting that other factors may be stronger determinants. The variability may have been affected by cover crop effects from the previous year, since the same organic field was used, and randomization was implemented. Few long-term cover crop studies have been published, but significantly higher invertebrate and nematode biomass in soil from maize plots intercropped with a legume cover crop over a 10-year period was measured compared to monoculture maize plots with or without mineral fertilizer application (30). Interestingly, Schutter et al. (31) found that season, not cover cropping system, was the most influential determinant of microbial community structure in vegetable fields with a 5-year history of winter cover crop rotations. The bacterial population dynamics that we observed in soil were specific to the temperature requirements of the two indicator taxa used. Both E. coli and L. innocua exhibited population declines with increasing time from inoculation. However, E. coli populations waned when temperatures dipped to <5°C and reappeared only sporadically in the spring and summer, while L. innocua persisted throughout the study. Jiang et al. (32) reported more-rapid declines in E. coli O157:H7 numbers in manured soil at 5°C than at 15°C and 21°C. Moreover, the carbon status of soil may interact with temperature, resulting in more detrimental cold and starvation stresses on E. coli O157:H7 (33). This raises the question of whether cover crop species that accrue carbon over time might favor temperature-sensitive bacteria. The failure of E. coli to resurge when temperatures warmed the following spring might be attributed to the significant die-off. Additionally, any recovery attained during the milder spring period might again be restricted by soaring temperatures in the summer. E. coli O157:H7-contaminated agricultural soil exhibited shorter survival duration in tropical soil than in temperate soil, partly attributed to temperature and moisture levels, combined with UV radiation (34). E. coli is not ubiquitous in agricultural environments, as it is typically introduced through manure or animal feces. E. coli detected on deer pellets grew at maximal rates at 20°C rather than at 4°C or 35°C, with minimal growth observed at 4°C (35); this is consistent with the results of E. coli survival studies in dairy manure (36) and cattle feces (37). Higher water content has also been associated with a decrease in microbial die-off rates in soil (38). Soil type greatly determines the soil water content. The extended survival of nonpathogenic E. coli and E. coli O157:H7 has been attributed to higher water availability in different soil types (39) and to moisture content (40), while the long-term survival of Listeria monocytogenes in soil has been attributed to soil texture and clay content (41, 42). Sandy soils, like the soil in our study, retain less water and have been associated with faster declines in E. coli (43). We also observed declines in the populations of both E. coli and L. innocua following moderate rain events, suggesting that bacteria were removed from upper soil layers through infiltration. These rain events were generally succeeded by population resurgence. On the other hand, heavier rain (>25 mm) did not appear to affect the populations, probably as a result of larger amounts of rainfall contributing to the formation of a soil surface seal, resulting in a reduction in water infiltration (44) and bacteria remaining in surface layers. Cover crop root systems are expected to improve water infiltration and soil water retention (45).

While mesophilic E. coli was outcompeted during the colder months, exhibiting very weak resurgence in the spring, L. innocua persisted throughout the winter, allowing it to prevail at high levels in the spring. Both soil and decaying vegetation can serve as reservoirs of Listeria spp., and the bacteria have resilience under harsh or nutrient-scarce environmental conditions (46, 47). Temperature has been shown to play an important role in Listeria survival in soil, and it has a competitive advantage at lower temperatures. During the course of this study, Arctic air carried by the jet stream brought record cold temperatures to the eastern United States. In Maryland, January 2014 (year 1) and February 2015 (year 2) were the 12th (48) and 6th (49) coldest on record, respectively. Previous research has shown that L. monocytogenes survives better at 5°C than at 15 to 20°C (50, 51), and it is able to survive in soil for extended periods at low temperatures (52, 53). The persistence of pathogenic Listeria spp. over the winter is a worrying prospect. Genes associated with nutrient acquisition mechanisms are overrepresented in L. monocytogenes, enabling the pathogen to generate energy from a wide range of substrates (54), and long-term persistence in stationary phase was reported not to compromise its ability to cause infection (55). L. monocytogenes is also frequently associated with manure (56) and the presence of wildlife (57), two common on-farm risk factors. Due to its recalcitrance to inactivation and adaptability to the soil niche, allowing longer time intervals between manure application and harvest might not be sufficient to reduce the likelihood of dissemination onto crops. However, employing strategies that improve soil microbial diversity, such as varying the plant genotypes within a field (reviewed in reference 58), may provide a biological barrier against the establishment of L. monocytogenes (59).

In the organic field, tillage, but not green manure type, appeared to correlate with larger bacterial populations in the soil. No differences among cover crop and bare-ground treatments were observed. As temperatures warm up, spring tillage and green manure incorporation may contribute to the resurgence of bacterial populations in the soil as a result of added carbon and enhanced soil porosity and oxygen diffusion (60). In our study, tillage did not diminish E. coli populations but appeared to be detrimental to L. innocua in the organic field. Subsequent recovery was eventually observed, although it is not possible to determine whether other factors, such as temperature, contributed to this effect, since a no-till control trial was not conducted. Conversely, in the transitional field, green manure type did indeed have an effect. E. coli was highest in the crimson clover-rye mixtures, and the bacterial population within this treatment was different than that in either cover crop (crimson clover or rye) in monoculture and bare-ground plots. The effects of green manure type on L. innocua were noted, with the hairy vetch-rye mixture harboring the smallest bacterial populations and rye the largest. In spite of the benefits, tillage events can cause temporary stress conditions for soil microbes, decreasing their ability to assimilate nutrients, altering microbial community structure, and increasing the potential for the loss of carbon and nitrogen from the soil and degradation of organic matter (61). The incorporation of green manure may temper these effects as a result of nutrient inputs and reduced soil compaction. Conversely, in a no-till setting, crop residues can serve as insulators, attenuating soil temperature fluctuations, conserving soil moisture, and building up organic matter and nutrients (62, 63). Soil organic matter has been reported to be correlated with a prolonged persistence of E. coli O157:H7 (40). There is therefore a need to assess the long-term effects of green manure on foodborne pathogen populations.

Although cover crops may have an impact on bacterial population dynamics, they were not the only or most influential driving force. The utilization of cover crops and green manures as possible biocontrol strategies against foodborne pathogens does not appear to be promising. More problematic is the potential of the cover crop rhizosphere or green manure to promote microbial growth, including that of foodborne pathogens, since prolonged presence in the soil increases the likelihood of transmission onto crops (64). This is an important consideration for farmers who produce crops that are considered high risk from a food safety standpoint, especially if their farm management program includes the use of multispecies cover crop cocktails in which high biomass is attained. E. coli and L. innocua were used in this study as surrogates for pathogenic enteric bacteria. Generic E. coli has been reported to behave in parallel with pathogenic strains (39) and with Salmonella enterica serotype Typhimurium (65) in the soil. L. innocua was used in this study as a surrogate for the pathogenic species L. monocytogenes, as the two species exhibit similar behaviors in soil, although L. innocua may be better adapted (66). Our findings therefore suggest that in regions with cold winters where die-off of enteric bacteria is expected, cover cropping may prolong the survival of mesophilic pathogenic bacteria, as seen in our study with E. coli in the hairy vetch-rye and rye plots in the first year. On the other hand, psychrotrophic Listeria spp. are already well adapted to persist in soil, and the benefits of a rhizosphere effect may be less significant in view of their robust competitiveness. Soil type is another important factor influencing bacterial infiltration and survival, with extended pathogen survival observed in subsurface loamy and clay soils (67). Future studies should evaluate how soil type interacts with cover crop rhizospheres to influence enteric pathogen population dynamics. Ultimately, preventing the introduction of enteric bacteria into production areas remains the most important step that growers can take to prevent them from becoming established there. The application of pathogen-free composted manure is an important recommendation for fresh-produce growers, although wildlife exclusion continues to be a challenge. The Food Safety Modernization Act (68) has deferred recommendations for soil amendments until such a time that more data can be acquired. Considering the important ecosystem services provided by cover crops and the benefits of organic fertilization on soil fertility and health, future studies should assess the short-term and long-term impacts of cover cropping on enteric bacterial population dynamics in various soil types, including pathogens which may be introduced to the soil via manure amendments or by wildlife.

Funding Statement

This work was supported by the U.S. Department of Agriculture’s National Institute of Food and Agriculture through the Organic Transitions Program under grant number 2014-51106-22090 to Shirley A. Micallef. Any opinions, findings, and conclusions expressed in this material are those of the authors and do not necessarily reflect the views of the USDA NIFA.