Stockpiling versus Composting: Effectiveness in Reducing Antibiotic-Resistant Bacteria and Resistance Genes in Beef Cattle Manure

ABSTRACT

Manure storage methods can affect the concentration and prevalence of antibiotic-resistant bacteria (ARB) and antibiotic resistance genes (ARGs) in cattle manure prior to land application. The objective of this study was to compare stockpiling and composting with respect to their effectiveness in reducing ARB and ARGs in beef cattle manure in a field-scale study. Field experiments were conducted in different seasons with different bulking agents for composting. For both the winter-spring cycle and the summer-fall cycle, ARB concentrations declined below the limit of quantification rapidly in both composting piles and stockpiles; however, ARB prevalence was significantly greater in the composting piles than in the stockpiles. This was likely due to the introduction of ARB from bulking agents. There was no significant change in ARG concentrations between initial and final concentrations for either manure storage treatment during the winter-spring cycle, but a significant reduction of the ARGs erm(B), tet(O), and tet(Q) over time was observed for both the composting pile and stockpile during the summer-fall cycle. Results from this study suggest that (i) bulking agent may be an important source of ARB and ARGs for composting; (ii) during cold months, the heterogeneity of the temperature profile in composting piles could result in poor ARG reduction; and (iii) during warm months, both stockpiling and composting can be effective in reducing ARG abundance.

IMPORTANCE Proper treatment of manure is essential to reduce the spread of antibiotic resistance and protect human health. Stockpiling and composting are two manure storage methods which can reduce antibiotic-resistant bacteria and resistance genes, although few field-scale studies have examined the relative efficiency of each method. This study examined the ability of both methods in both winter-spring and summer-fall cycles, while also accounting for heterogeneity within field-scale manure piles. This study determined that bulking agents used in composting could contribute antibiotic-resistant bacteria and resistance genes. Additionally, seasonal variation could hinder the efficacy of composting in colder months due to heterogeneity in temperature within the pile; however, in warmer months, either method of manure storage could be effective in reducing the spread of antibiotic resistance.

INTRODUCTION

The spread of antibiotic resistance is of growing public health concern, as many bacterial pathogens have developed resistance to routine antibiotics (1, 2) and drugs of last resort, such as carbapenems (3, 4). The spread of antibiotic resistance is partially attributed to the use of antibiotics in livestock operations, which accounts for up to 80% by mass of total antibiotic usage in the United States (5, 6). Administration of antibiotics to livestock can increase antibiotic-resistant bacteria (ARB) as well as antibiotic resistance genes (ARGs) in livestock wastes (7, 8). Additionally, livestock wastes can be a reservoir for zoonotic pathogens such as Escherichia coli O157:H7 (9, 10), Salmonella enterica serovar Typhimurium (11), and Listeria and Campylobacter spp. (12). Consequently, the use of untreated livestock manure for land application may facilitate the spread of ARB, ARGs, and pathogens to soils and water (13–15). Soil and water receiving these microbial agents could facilitate their transmissions to human through contamination of food crops and direct exposure to the contaminated environment (16).

Hence, it is important that proper manure management strategies are used prior to land application. One common practice for manure storage is stockpiling (17, 18), through which pen surface materials are piled and the piles are allowed to sit and age until land application. Manure stockpiles can sometimes heat up initially; however, they may not reach a sufficiently high temperature or maintain the elevated temperature for a sufficient length of time to reduce ARG/ARGs. Hence, stockpiling may not be sufficient treatment to kill off zoonotic pathogens and reduce ARB/ARGs (10).

Composting is an alternative to stockpiling. According to the Food and Drug Administration Produce Safety Rule for application of soil amendments of animal origin, static composting must maintain aerobic conditions and temperatures of 55°C for 3 consecutive days, while turned composting must maintain 55°C for 15 cumulative days, with a minimum of five turning events (19). Similarly, Environmental Protection Agency guidelines require biosolids from municipal wastewater treatment facilities to be composted at 55°C and above for 3 days (20, 21). The heat generated during composting is derived from aerobic microbial activities. A bulking agent (e.g., corn stalk residues, wood chips, or sawdust) is commonly added to manure to increase oxygen penetration and meet microbial nutrient needs (22).

While composting is often considered more reliable than stockpiling to reduce pathogens in manure, it is also considerably more time-consuming, costly, and labor-intensive to establish and operate (23–25). The efficacy of composting for reducing ARB and ARGs has been investigated, mostly with lab-scale setups and some with field-scale piles (26). Although lab-scale reactors are simple to establish, they cannot adequately simulate full-scale composting piles. Hence, conclusions from lab-scale composting studies may not be directly applicable to field-scale composting piles. Field-scale studies have reported inconsistent results on the efficacy of composting on ARB/ARG reduction (26). Some field-scale studies have reported consistent decreases in ARB/ARG abundance during composting (27–31), while others have noted inconsistent effects (i.e., some ARGs decreased while others persisted or even increased) (32–35). This is not surprising because of the variability in manure characteristics, choice of bulking agents, ambient temperature, and pile operation, which could all affect the inactivation of ARB and reduction of ARGs. In addition, owing to the heterogeneity in full-scale manure piles, the sampling strategy employed to collect samples from manure piles can also affect the conclusions. To comprehensively assess the efficacy of composting in reducing ARB and ARGs in livestock manure, more full-scale studies with systematic sample strategies, to account for manure pile heterogeneity, are needed.

The objective of this study was to investigate the effectiveness of composting in reducing ARB and ARGs in beef cattle manure using field-scale experiments. This required an experimental design to account for the suspected heterogeneity inside the stockpiles and composting piles. Both the stockpile and composting piles were operated for the length of time required for two complete composting cycles. One cycle was operated during the winter-spring and one cycle was operated during the summer-fall, allowing for assessment of seasonal differences in full-scale systems. To better account for the piles, samples were collected at two or three profiles along each pile and at three depths per profile. Additionally, detailed temperature profile data were also documented from each pile. These measures were used to facilitate a more comprehensive and objective comparison between the stockpiling and composting methods.

RESULTS

The winter-spring cycle.

(i) Temperature. After pile establishment (16 December 2017), the interior temperatures of all four piles, two composting piles and two stockpiles, quickly exceeded 55°C (see Table S2 in the supplemental material). The two stockpiles exhibited temperatures above 55°C for about 15 days, mostly at layer 4 (138 cm from the base of the pile) of the piles (Fig. 1). In contrast, the two composting piles exhibited temperatures above 55°C in more layers and over a longer period (Fig. S2). For example, composting pile 2 retained temperatures exceeding 55°C in layers 2 to 4 for about 20 days. After the first turning (29 January 2018), both composting piles achieved temperatures exceeding 55°C in layers 2 to 4 for at least 5 more days. After the second turning (2 April 2018), both composting piles exhibited temperatures above 55°C in at least two layers for 9 days. The topmost layer in both composting piles stayed at temperatures above 55°C for 15 days (Table S2). The base of the piles never reached the desired temperature of 55°C, as the ground was frozen during the winter months.

FIG 1.

(A) Placement of thermocouples and sampling locations within the composting piles and stockpiles. Black dots represent thermocouple placement, and red crosses represent sampling locations. (B) Photograph of one of the composting piles from the winter-spring cycle.

(ii) Bacterial concentrations and prevalence. Initial concentrations of total E. coli and enterococci were higher in the pen scrapings than in the corn stalk residues prior to incorporation (Table 1). However, the corn stalk residues had the only quantifiable concentration of azithromycin-resistant (azithromycin^R) E. coli, which was not detected in the initial pen scrapings (Table 1). After the initial establishment of manure piles, the concentrations of total E. coli (Fig. S2) and total enterococci (Fig. S3) declined rapidly in the stockpiles. In all three depths of the stockpiles, total E. coli concentrations were undetectable after day 22, and total enterococcus concentrations were undetectable after day 43. In contrast, both bacteria continued to appear sporadically in the composting piles, particularly after each turning event (Fig. S2 and S3). Antibiotic-resistant E. coli and enterococci fell below the limit of quantification after the day 0 sampling event for all piles.

TABLE 1.

Initial concentrations of total and antibiotic-resistant E. coli and enterococci in open feedlot pen scrapings and in the bulking agent for both cycles of the study

Cycle and sample type	Initial concn (log CFU g−1 [wet wt])
	Total E. coli	AzithromycinR E. coli	TetracyclineR E. coli	Total enterococci	TylosinR enterococci	TetracyclineR enterococci
Winter-spring
Pen scrapings	7.63	BDLa	BDL	11.63	1.38	2.63
Corn stalk residues	4.15	4.30	BDL	4.34	BDL	3.30
Summer-fall
Pen scrapings	BDL	BDL	BDL	3.78	BDL	BDL
Rye silage	BDL	BDL	BDL	4.93	BDL	BDL

^aBDL, below detection limit.

Analysis of variance (ANOVA) revealed that there was a significant main effect of manure storage method on total E. coli (P < 0.001) and total enterococcus (P < 0.001) (Table 2) concentrations. Contrary to expectation, higher concentrations of both bacteria were observed in the composting piles than in the stockpiles (Table 2). ANOVA also revealed a significant effect of depth for total E. coli and total enterococci (P = 0.008 and P = 0.001, respectively), with the highest concentrations in the 2.5-cm-depth samples. Post hoc analyses revealed that this was true only for composting piles.

TABLE 2.

Means and P values for storage method and depth for bacterial concentrations and bacterial prevalence in the winter-spring cycle^a

Parameter	Concn (log10 CFU g−1)b		Prevalence (%)
	Total E. coli	Total enterococci	Total E. coli	AzithromycinR E. coli	TetracyclineR E. coli	Total enterococci	TylosinR enterococci	TetracyclineR enterococci
Storage methodc,d
Composting	1.45 A	1.81 A	58 A	5 A	10 A	96 A	82 A	85 A
Stockpiling	0.22 B	0.32 B	18 B	1 B	5 B	62 B	31 B	27 B
Depth from surface (cm)c,d
2.5	1.17 A	1.54 A	44	2 A	13 A	92 A	69 A	69 A
45.7	0.76 B	0.98 B	38	6 B	6 B	86 A	59 A	58 A
76.2	0.58 B	0.68 B	32	2 A	5 B	60 B	42 B	42 B
P value
Method	<0.001	<0.001	0.002	<0.001	0.031	<0.001	<0.001	<0.001
Depth	0.008	0.001	0.113	0.016	<0.001	<0.001	0.003	<0.001

^aStatistics for interaction terms can be found in Table S3.

^bAntibiotic-resistant E. coli and enterococci were not quantifiable.

^cValues followed by a letter combination sharing one or more letters are not statistically different at the P value of <0.05 based on Tukey’s post hoc test.

^dValues are treatment averages which were calculated based on all data for one particular treatment level.

Similarly, ANOVA results showed a significant effect of manure storage method on the prevalence of all bacterial populations tested. The frequency of detection was significantly greater for both E. coli and enterococci, total or resistant, in composting piles than in stockpiles (P ≤ 0.031 for all cases [Table 2]; see Table S3 for more detailed statistics). The ANOVA also revealed a significant effect for depth on all but one bacterial population (P ≤ 0.016 for all analyses except for total E. coli [Table 2 and Table S3]). In general, bacterial prevalence was significantly greater in 2.5-cm samples than in 76.2-cm samples.

The prevalence data for the two manure storage methods are illustrated in Fig. 2 and 3. The prevalence of both E. coli and enterococci, total and resistant, was consistently greater in composting piles than in stockpiles. Further, the detection frequency within the composting piles tended to increase followed turning events on days 49 and 112 (Fig. 2 and 3). Noticeably, at the end of the 140-day experiment, the total and resistant E. coli organisms were no longer detectable in the stockpiles but were still detected frequently in the composting piles.

FIG 2.

Prevalence at three depths on each sampling event of total E. coli (A), azithromycin^R E. coli (B), and tetracycline^R E. coli (C) in stockpiles and total E. coli (D), azithromycin^R E. coli (E), and tetracycline^R E. coli (F) in the composting piles during the winter-spring cycle. Black lines (days 49 and 112) indicate dates on which composting piles were turned and samples were taken immediately after. Error bars represent standard errors (n = 4 profiles per sampling event).

FIG 3.

Prevalence at three depths on each sampling event of total enterococci (A), tylosin^R enterococci (B), and tetracycline^R enterococci (C) in stockpiles and total enterococci (D), tylosin^R enterococci (E), and tetracycline^R enterococci (F) in the composting piles during the winter-spring cycle. Black lines (days 49 and 112) indicate dates on which composting piles were turned and samples were taken immediately after. Error bars represent standard errors (n = 4 profiles per sampling event).

(iii) ARG concentrations. Based on the ANOVA, there was no significant effect for manure storage methods for most of the genes tested, except for intI1 (P = 0.003 [Table 3]). The intI1 concentration was greater in the composting piles than in the stockpiles (Table 3 and Table S4). A significant effect of depth was also observed only for the 16S rRNA gene (P = 0.045) and intI1 (P = 0.005). Post hoc tests indicate that this effect was driven by significantly greater concentrations at the 2.5-cm depth than at the 76.2-cm depth in stockpiles for the 16S rRNA gene (P = 0.035) and in both stockpiling and composting piles for intI1 (P = 0.045 and 0.049, respectively [Fig. 4]). ANOVA revealed a significant time effect on the concentrations of all ARGs examined following the first turning on day 49 (29 January 2018) (Table S4). There was an increase in concentration for most ARGs, followed by a decrease of approximately 1 order of magnitude for the subsequent sampling event (Fig. 5).

TABLE 3.

Means and P values for storage method and depth for ARG concentrations in the winter-spring cycle^a

Parameter	Concn (log10 CN g−1)
	16S rRNA gene	intI1	erm(B)	erm(F)	tet(O)	tet(Q)
Storage methodb,c
Composting	9.62	7.77 A	4.93	6.88	4.33	6.01
Stockpiling	9.31	7.37 B	4.53	6.13	3.95	5.55
Depth from surface (cm)b,c
2.5	9.60 A	7.74 A	4.72	6.85	4.19	5.94
45.7	9.41 B	7.60 B	4.78	6.42	4.17	5.90
76.2	9.38 AB	7.36 B	4.70	6.24	4.06	5.49
P value
Method	0.262	0.003	0.333	0.245	0.296	0.326
Depth	0.045	0.005	0.721	0.225	0.909	0.117
Time	0.023	0.007	0.027	0.022	0.004	0.009

^aStatistics for interaction terms can be found in Table S4.

^bValues followed by a letter combination sharing one or more letters are not statistically different at the P value of <0.05 based on Tukey’s post hoc test.

^cValues are treatment averages which were calculated based on all data for one particular treatment level.

FIG 4.

Concentrations at three depths on each sampling event of the 16S rRNA gene (A) and intI1 (B) in a stockpile and the 16S rRNA gene (C) and intI1 (D) in a composting pile during the winter-spring cycle. Black lines (days 49 and 112) indicate dates on which composting piles were turned and samples were taken immediately after. Error bars represent standard deviations (n = 2 profiles per sampling event).

FIG 5.

Concentrations of ARGs for a stockpile (A) and a composting pile (B) over time during the winter-spring cycle. Black lines (days 49 and 112) indicate dates on which compost piles were turned and samples were taken immediately after. Error bars represent standard errors (n = 6, 2 profiles × 3 depths per sampling event).

Summer-fall cycle.

(i) Temperature. Temperatures for nearly all measured locations within the composting pile were greater than 55°C for all 47 days of the summer-fall cycle (8 August 2018 to 24 September 2018 [Table S5]). Consequently, the composting pile was not turned during this cycle of the experiment. Layers 3 and 4 within the stockpile consistently exceeded 55°C for about 27 days (Table S5). However, temperatures measured near the bottom of the stockpile never exceeded 55°C (Table S5).

(ii) Bacterial concentrations and prevalence. Pen surface materials for the summer-fall cycle of the study had considerably lower bacterial concentrations than for the winter-spring cycle, with only total enterococci being detected above the limit of quantification (Table 1). The rye silage used as a bulking agent had no detectable ARB or total E. coli but had total enterococcus concentrations greater than those of the pen surface materials (Table 1). Similar to the case with the winter-spring cycle, resistant E. coli and enterococci were consistently below the limit of quantification throughout the study, and quantifiable total E. coli and enterococcus concentrations occurred sporadically in the stockpile and composting pile.

ANOVA results showed that the manure storage method had a significant impact on the concentration of total E. coli (P = 0.019 [Table 4]). Similar to what was found in the winter-spring cycle, total E. coli concentrations were higher in the composting pile than in the stockpile. Also, the manure storage method had a significant impact on the prevalence of resistant E. coli, total enterococci, and tetracycline^R enterococci. For these four populations, there were significantly greater prevalences in the composting pile than in the stockpile (Fig. 6 and 7 and Table 4; see Table S6 for more detailed statistics). Depth did not appear to significantly impact the concentrations or the prevalence of the tested bacterial populations.

TABLE 4.

Means and P values for storage method and depth for bacterial concentrations and bacterial prevalence in the summer-fall cycle^a

Parameter	Concn (log10 CFU g−1)b		Prevalence (%)
	Total E. coli	Total enterococci	Total E. coli	AzithromycinR E. coli	TetracyclineR E. coli	Total enterococci	TylosinR enterococci	TetracyclineR enterococci
Storage methodc,d
Composting	0.66 A	0.85	54	18 A	13 A	87 A	23	36 A
Stockpiling	0.52 B	0.82	49	3 B	0 B	67 B	19	17 B
Depth from surface (cm)c,d
2.5	0.80	0.98	61	15	8	88	29	33
45.7	0.61	0.74	49	8	4	80	21	29
76.2	0.33	0.78	45	8	8	64	13	17
P value
Method	0.019	0.518	0.462	0.001	<0.001	<0.001	0.060	0.002
Depth	0.265	0.594	0.477	0.198	0.683	0.120	0.198	0.256

^aStatistics for interaction terms can be found in Table S6.

^bAntibiotic-resistant forms of E. coli and enterococci were not quantifiable.

^cValues followed by a letter combination sharing one or more letters are not statistically different at the P value of <0.05 based on Tukey’s post hoc test.

^dValues are treatment averages which were calculated based on all data for one particular treatment level.

FIG 6.

Prevalence at three depths on each sampling event of total E. coli (A), azithromycin^R E. coli (B), and tetracycline^R E. coli (C) in the stockpile and total E. coli (D), azithromycin^R E. coli (E), and tetracycline^R E. coli (F) in the composting pile during the summer-fall cycle. Error bars represent standard errors (n = 3 profiles per sampling event).

FIG 7.

Prevalence at three depths on each sampling event of total enterococci (A), tylosin^R enterococci (B), and tetracycline^R enterococci (C) in the stockpile and total enterococci (D), tylosin^R enterococci (E), and tetracycline^R enterococci (F) in the composting pile during the summer-fall cycle. Error bars represent standard errors (n = 3 profiles per sampling event).

(iii) ARG concentrations. ANOVA results indicate that the manure storage method had no significant effect on the genes tested, except erm(F). The erm(F) gene was significantly lower in the composting pile than in the stockpile (Table 5 and Table S7). A significant main effect for depth was detected for tet(Q) (P = 0.005). The tet(Q) gene concentration was significantly greater at the 76.2-cm depth than at the 45.7-cm and 2.5-cm depths (P = 0.009 and 0.007, respectively [Fig. 8]). Also, ANOVA results indicated a significant time effect for the four ARGs (Table S7). For erm(B), tet(O), and tet(Q), this effect was driven by an overall decline in concentrations in both the stockpile and composting pile (Fig. 9), while for erm(F), no overall decline was noted and this effect was driven by an increase in concentrations on 27 August 2018 (day 19).

TABLE 5.

Means and P values for storage method and depth for ARG concentrations in the winter-spring cycle^a

Parameter	Concn (log10 CN g−1)
	16S rRNA	intI1	erm(B)	erm(F)	tet(O)	tet(Q)
Storage methodb,c
Composting	9.35	7.15	4.31	6.25 A	4.66	5.33
Stockpiling	9.52	7.39	4.78	6.98 B	4.91	5.70
Depth from surface (cm)b,c
2.5	9.46	7.38	4.54	6.87	4.55	5.25 A
45.7	9.52	7.40	4.45	6.39	4.69	5.31 A
76.2	9.31	7.04	4.65	6.58	5.11	5.97 C
P value
Method	0.367	0.508	0.162	0.013	0.447	0.472
Depth	0.106	0.083	0.377	0.053	0.106	0.005
Time	0.159	0.393	<0.001	0.002	<0.001	<0.001

^aStatistics for interaction terms can be found in Table S7.

^bValues followed by a letter combination sharing one or more letters are not statistically different at the P value of <0.05 based on Tukey’s post hoc test.

^cValues are treatment averages which were calculated based on all data for one particular treatment level.

FIG 8.

Concentrations of tet(Q) at each sampling depth from two profiles on each sampling day of the summer-fall cycle in the stockpile (A) and the composting pile (B). Error bars represent standard deviations (n = 2 profiles per sampling event).

FIG 9.

Concentrations of the 16S rRNA gene, intI1, and three ARGs from two profiles for each manure storage method over time during the summer-fall cycle in the stockpile (A) and the composting pile (B). Error bars represent standard errors (n = 6; 2 profiles × 3 depths per sampling event).

DISCUSSION

A review of the available literature showed a limited number of full-scale studies evaluating the effect of composting of material removed from open-lot beef feedlots on ARB and ARGs. Some studies reported effective reduction of ARB and ARGs during composting (28, 36, 37), while others reported inconsistent or contrary results (32–35). Among previous field-scale studies, differences in manure composition (e.g., caused by species, feed, etc.) and composting process (e.g., bulking agents) are likely responsible for the seemingly contradictory findings. More importantly, the heterogeneity of field-scale manure piles may or may not be sufficiently accounted for by sampling strategies. Therefore, this study was designed to account for the heterogeneities by including two complete composting cycles using different manure, two distinct seasons, and monitoring/sampling multiple locations for temperature and ARB/ARGs throughout the manure piles.

Effects on temperature.

As expected, the internal temperatures of composting piles were higher than those of stockpiles (Tables S2 and S5). Also, there was less heterogeneity in the temperature profiles of both composting and stockpiles in the summer-fall cycle than in the winter-spring cycle. This was due to elevated atmospheric and ground temperatures. The mean monthly ambient temperatures were −2.8°C, −4.7°C, −5.3°C, 3.9°C, and 5.8°C from December to April during the winter-spring cycle and 22.5°C and 19.2°C in August and September during the summer-fall cycle. In the summer-fall cycle, temperatures exceeding 55°C were achieved nearly everywhere in the composting pile. In contrast, in the winter-spring cycle temperatures exceeding 55°C were reached only in upper portion of the composting piles due to frozen ground.

Effects on bacteria.

Our data suggest that certain bulking agents could introduce additional ARB to composting piles and that these elevated ARB concentrations could persist. Compared to their stockpiling counterparts, the composting piles in both cycles displayed greater prevalence of ARB (Tables 2 and 4). During the winter-spring cycle, the corn stalk residues used as the bulking agent contained high concentrations of ARB, including azithromycin^R E. coli, which was absent in pen scrapings (Table 1). A previous survey showed ARB are commonly found on corn stalk residues, with ARB detected in 54% of corn stalk residues surveyed in Nebraska (38). Unlike the corn stalk residues used in the winter-spring round, the bulking agent rye silage used in the summer-fall cycle contained no quantifiable ARB (Table 1). However, post hoc analyses found significantly greater prevalence of azithromycin^R E. coli, tetracycline^R E. coli, total enterococci, and tetracycline^R enterococci in the composting pile than in the stockpile (Table 4). This is likely attributable to rye silage. As the bulking agents were mixed with pen scrapings at a 40:60 ratio, they could represent a considerable source of ARB to the composting piles. Hence, choosing a bulking agent free of ARB could be important to achieving ARB control in composting piles. It is noted that in the summer-fall round, both pen scrapings and rye silage initially had E. coli concentrations below the limit of detection. However, quantifiable E. coli concentrations were later found in both piles, suggesting the possibility of growth of E. coli in portions of the manure piles during both stockpiling and composting.

Turning events appeared to cause transient increases in ARB concentration in composting piles. Turning could stimulate microbial activities by redistributing nutrients and replenishing oxygen inside manure piles, sustaining the heating of composting piles (39). During the winter-spring cycle with the low atmospheric temperature and frozen ground, portions of the composting pile never exceeded 55°C. Instead, temperatures favorable to bacterial growth (∼35 to 45°C) (40, 41) occurred in the lower section of the piles sitting on frozen ground. In contrast, the base of the stockpiles remained at very low temperatures throughout the experiment. During turning events, materials that previously resided in portions where temperatures favorable to bacterial growth occurred were mixed with materials from other portions of the manure piles and served as bacterial seed. This may explain why there were transient increases in bacterial concentration (Fig. S2 and S3) and prevalence (Fig. 2 and 3) after turning.

In order to identify common trends between the relationship of temperature and prevalence of indicator bacteria under the two treatment regimes, we combined the data from the winter-spring round and the summer-fall round of experiments. When all data points were pooled, it was found that there were consistently higher prevalences of total and resistant E. coli and enterococci under composting treatment than under stockpiling treatment (Fig. S4A and C). When only data points with measured temperatures at or above 55°C were considered, the distinctions in prevalence between composting and stockpiling become much less pronounced (Fig. S4B and D). These findings suggest that temperature above 55°C is a reliable predictor for the inactivation of total and resistant indicator bacteria regardless of treatment regime. Furthermore, when bulking agents potentially contribute to the total and resistant indicator bacteria, the portions of composting piles with temperatures below 55°C become the “hot spots” of these microbes.

Effects on ARGs.

No obvious advantage of composting was observed with respect to ARG reduction in either cycle of this study. Neither stockpiling nor composting showed a significant reduction in ARGs when their final concentrations were compared with the initial concentrations during the winter-spring cycle according to post hoc analyses, despite fluctuating ARG concentrations over time (Table 3). Because the ARG reduction performance was poor for both stockpiling and composting, there was no significant difference between the two methods (Table 3 and Table S4). In contrast, during the summer-fall cycle, post hoc analyses showed that the final concentrations of erm(B) and tet(O) were significantly lower than the initial concentrations in both stockpiles (P = 0.020 and 0.007, respectively) and composting piles (P = 0.007 and 0.038, respectively). Because a significant reduction occurred in both stockpiles and composting piles, the difference between the two storage methods was not significant (Table 5).

The fate of ARGs in manure composting piles is likely linked to the fate of their bacterial hosts, based on their thermotolerance (28, 36, 42). Composting with sawdust (30) and red mud (29) as bulking agents could result in reduction of ARGs by stimulating the growth and activity of certain microbial populations in composting piles (43, 44). Inconsistent reduction in ARGs and alterations in the microbial community were noted in a field-scale study composting cattle manure with straw as a bulking agent (32). In that 2016 study by Qian et al., the ARGs tet(C) and tet(X) increased, while tet(Q), tet(M), and tet(W) decreased, as different bacterial community profiles were found to be associated with changes in the ARG profile as a factor of composting time. Another field-scale composting study in which hyperthermophilic temperatures (>90°C) were reached similarly found significant reductions in tet genes, associating the reduction with the removal of thermotolerant bacterial species and associated mobile genetic elements (27). Several studies have also linked ARG expression with changes in microbial community composition during the composting process. For example, Firmicutes were the most transcriptionally active ARG carriers during the mesophilic phase (∼40°C) of composting, while Firmicutes, Actinobacteria, Bacteroides, and Proteobacteria were the most transcriptionally active ARG carriers during the thermophilic phase (>60°C) of the composting, in one study (45). Consequently, despite notable declines in less thermotolerant bacterial taxa, ARG abundance can remain the same as antibiotic-resistant thermotolerant taxa thrive or obtain ARGs via horizontal gene transfer under thermophilic conditions (46).

Compared to stockpiling, composting did not consistently reduce ARB or ARGs in the open feedlot pen scrapings in this study. The use of a bulking agent in composting piles appeared to increase ARB concentrations and prevalence due to the introduction of additional ARB from bulking agents. The heterogeneity in temperature profiles, caused by low ambient air temperature and frozen ground in winter and spring months, also contributed to increased ARB prevalence in composting piles by creating temperature gradients that were favorable to bacterial growth. Composting did not exhibit superior performance on ARG reduction over stockpiling, as no ARG was significantly reduced under either method in the winter-spring cycle and was significantly reduced under both methods in the summer-fall cycle. Results from this study suggest that (i) bulking agent can be an important source of ARGs for composting; (ii) during cold months, the heterogeneity of the temperature profile in composting piles could result in poor ARG reduction; and (iii) during warm months, both stockpiling and composting can be effective in reducing ARG abundance.

MATERIALS AND METHODS

Setup and sampling of composting and stockpiles.

(i) Winter-spring. Two composting piles and two stockpiles were established at the USDA U.S. Meat Animal Research Center (MARC) near Clay Center, NE. The stockpiles consisted of pen scrapings from beef cattle feedlot pens, while the composting piles consisted of a 60:40 ratio of pen scrapings to ground corn stalks, which were used as the bulking agent. This ratio was determined based on literature values for the carbon and nitrogen levels in cattle manure and corn stalk residues to achieve a final C/N ratio of 30:1 in compost piles (47). All four piles were approximately 6.1 m long, 3.0 m wide, and 1.8 m high. All four piles were set up on 11 December 2017 and operated until 30 April 2018. Composting piles were turned by physically moving the piles to an adjacent site 3 m to the side using a Bobcat front loader after day 49 (29 January 2018) and day 112 (2 April 2018), when the interior temperature dropped and remained below 50°C. Within each composting and stockpile, eight type-T thermocouples (Omega Engineering, Inc., Norwalk, CT) were placed along four horizontal layers within a profile (Fig. 1), with vertical distances 30.5, 45.8, 91.5, and 138 cm from the ground, to record temperature hourly using an OM-CP-OCTTemp A 8-channel multiplexer/data logger (Omega Engineering, Inc.).

Samples were collected from the four piles after the initial establishment and on days 7, 22, 28, 35, 43, 49, 56, 70, 84, 98, 112, 119, and 140. A grab sample of the corn stalk, prior to being incorporated into the composting piles, was collected for ARB analysis. To account for the potential heterogeneity along the manure piles, samples were collected from two profiles per pile at all sampling intervals. At each sampling interval, the sampling profiles were moved 0.15 m away from the immediate past sampling event. At each sampling profile, samples were collected from three depths from pile surface, 2.5 cm, 45.7 cm, and 76.2 cm, into the center of the pile (Fig. 1). Hence, six samples (2 profiles × 3 depths) were collected per pile per sampling event. Approximately 50 g of material was obtained at each depth. Samples at the 2.5-cm depth were collected via grab sampling, while those at the 45.7- and 76.2-cm depths were collected using an ethanol-sterilized push probe. Only the bottom section of manure in the push probe was recovered to avoid bacterial contamination from the superficial layer. All samples were placed in sterile 50-ml centrifuge tubes, stored on ice, and transported to the environmental engineering lab at the University of Nebraska—Lincoln (UNL) within 2 h of collection.

(ii) Summer-fall. During the second cycle of study, a composting pile and a stockpile were established and operated at MARC from 8 August 2018 through 24 September 2018. The stockpile similarly consisted of pen scrapings, while the composting pile consisted of a 60:40 ratio of pen scraping and rye silage, used as the bulking agent. Pile dimensions and thermocouple placement were the same as for the first cycle. The interior temperature of the composting pile never dropped below 50°C; hence, the composting pile was not turned during the study period.

The sampling procedure for the winter-spring cycle was largely adopted for the summer-fall cycle, with some modifications. Samples were collected after initial pile establishment and at days 5, 19, 33, and 47. Instead of two, three profiles were sampled per pile per sampling event. Within each sampling profile, samples from the same three depths as the first cycle were collected. A total of nine samples (3 profiles × 3 depths) were collected per pile per sampling event. A sample of rye silage was taken for ARB analysis prior to incorporation with pen scrapings. All samples were placed in polyethylene bags, stored on ice, and transported to the UNL environmental engineering lab within 2 h of collection.

ARB enumeration and detection.

All samples were assessed for abundance and presence of total culturable E. coli and enterococci, as well as culturable forms of both bacteria with resistance to a macrolide antibiotic (azithromycin and tylosin, respectively) and tetracycline. Tetracycline and tylosin are veterinary antibiotics commonly used on beef cattle and were used at the MARC facility (48–50). Tylosin and azithromycin both belong to the macrolide antibiotic family. Because E. coli is intrinsically resistant to tylosin, azithromycin was used to screen for macrolide-resistant E. coli (51). All solid samples were suspended 1:10 (wt/vol) in phosphate-buffered tryptic soy broth (Becton, Dickinson, Sparks, MD; TSB-PO₄), shaken vigorously for 2 min to dislocate bacteria associated with the samples, and plated for enumeration. E. coli organisms were enumerated on MacConkey agar (Becton, Dickinson) with no antibiotic, 20 mg liter⁻¹ of azithromycin, or 32 mg liter⁻¹ of tetracycline, while enterococci were quantified on Slanetz-Bartley agar (Oxoid, Hampshire, England) with no antibiotics, 32 mg liter⁻¹ of tylosin, or 32 mg liter⁻¹ of tetracycline (48–50). Concentrations of antibiotics were selected based on previously described clinical breakpoints and results with cattle manure (51, 52). An Eddy Jet 2 spiral plater (IUL, S.A., Barcelona, Spain) was used for plating. MacConkey plates were incubated at 37°C for 24 h, and Slanetz-Bartley plates were incubated at 44°C for 48 h. Quantifiable results are reported as CFU per gram of composted or stockpiled materials. Colony counting was performed using spiral counting grids provided by IUL, S.A., made specifically for the Eddy Jet 2 as per the instrument instruction manual, with a limit of quantification of 1 CFU 100 μl⁻¹ of suspension.

To assess the presence/absence of each bacterial target, all samples were subjected to a 24-h enrichment step for prevalence assessment. E. coli organisms were enriched in TSB-PO₄ and enterococci were enriched in Enterococcosel broth (Becton, Dickinson, Sparks, MD), both at 37°C for 24 h, prior to the same plating and incubation procedures as mentioned above.

To verify that colonies grown on MacConkey agar were E. coli, 20 isolates from one sampling event were selected and incubated in nutrient broth with 4-methylumbelliferyl-β-d-glucuronide (Difco, Sparks, MD). All were confirmed to be E. coli according to EPA method 9221F (53). An additional 426 colonies were also verified as E. coli via matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS; Shimadzu, Kyoto, Japan).

Propidium monoazide DNA extraction.

A subset of samples was chosen for ARG analyses from one composting and one stockpile in the winter-spring cycle (i.e., days 0, 7, 22, 35, 49, 70, 112, and 140) and from one composting and one stockpile in the summer-fall cycle (i.e., days 0, 5, 19, 33, and 47). A modified DNA extraction protocol involving propidium monoazide (PMA) was optimized for composted and stockpiled materials (Fig. S1) and employed in this study.

The modified DNA extraction procedure involved weighing out 0.1 g of sample, suspending it in 0.9 ml of autoclaved ultrapure water, and vortexing the suspension for 30 s. Fifteen microliters of 20 mM PMA was then added to the suspension for a final PMA concentration of 300 μM. Samples were then incubated in the dark for 5 min with shaking prior to being transferred to sterile Pyrex petri dishes. Petri dishes were placed on ice and exposed to a 500-W halogen light 15 cm away for 30 min (54, 55). The ice was replaced every 10 min. Following light exposure, the suspension was transferred to a PowerBead tube and DNA extraction was performed using a DNeasy PowerLyzer PowerSoil kit (Qiagen, Hilden, Germany) as per the manufacturer’s instructions.

ARG quantification.

Quantitative PCR (qPCR) was performed to quantify the 16S rRNA gene (56) and the intI1 gene (57), as well as macrolide resistance genes erm(B) and erm(F) (58) and tetracycline resistance genes tet(O) and tet(Q) (59). These genes were selected because of their prevalence in agricultural environments and their previous detection in manure obtained from MARC (58, 60–62). Each reaction consisted of 2× KiCqStart universal SYBR green ReadyMix (Sigma-Aldrich, St. Louis, MO) and final primer concentrations of 0.3 μM for genes intI1 and tet(Q), 0.375 μM for the 16S rRNA gene, 0.4 μM for erm(B), and 0.5 μM for genes erm(F) and tet(O). Reactions were carried out in duplicates, including no-template controls, on an Eppendorf Realplex² thermocycler (Eppendorf, Hamburg, Germany). Cycling conditions for each gene are presented in Table S1. Standard curves were run on all plates for all qPCR assays and were constructed using synthesized plasmid DNA (pIDTSMART with ampicillin resistance; Integrated DNA Technologies, Coralville, IA). DNA used for the standard curve was serially diluted in PCR-grade reagent water (Sigma-Aldrich). All qPCR runs had an efficiency between 90% and 105%, with an R² of >0.95. Results were reported as copy number (CN) per gram (wet weight) of manure or compost.

Statistical analyses.

For results from culture-based analyses, ARB were detected at quantifiable concentrations only immediately following pile establishment in both cycles and were therefore not included in the statistical analyses. With regard to bacterial prevalence (presence/absence), a split-plot ANOVA was conducted with manure storage method (i.e., composting versus stockpiling) as the main treatment factor and depth as a repeated-measure factor. Time was not included as a repeated-measure factor due to sporadic detection, and consequently, each time point was considered a replicate for storage method and depth. The response variable was the total number of detections of total and resistant E. coli and enterococci over the course of the study. Tukey’s post hoc test was performed if a significant effect was detected.

For results from qPCR analyses, split/split-plot ANOVA was conducted with manure storage method (i.e., composting versus stockpiling) as the main treatment factor and both depth and time as repeated measures providing a split-plot-in-time type of analysis. Log₁₀-transformed gene concentrations were response variables. For both cycles, gene analyses were conducted on two profiles from each manure pile and these profiles were treated as pseudoreplicates. All statistical analyses were performed in Statistica (Dell, Tulsa, OK) v.12, and results were considered significant at the level of 0.05.