This article was downloaded by: [University of Sussex Library] On: 09 February 2015, At: 13:54 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

British Poultry Science Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/cbps20

E-nose identification of Salmonella enterica in poultry manure a

a

b

c

d

Ü. Kizil , L. Genç , T. T. Genç , S. Rahman & M. L. Khaitsa a

Department of Agricultural Structures and Irrigation, Agricultural Sensor and Remote Sensing Laboratory, Çanakkale Onsekiz Mart University, Çanakkale, Turkey b

Department of Biology, Çanakkale Onsekiz Mart University, Çanakkale, Turkey

c

Department of Agricultural and Biosystems Engineering, North Dakota State University, North Dakota, USA d

Department of Pathobiology and Population Medicine, Mississippi State University, Mississippi State, USA Accepted author version posted online: 03 Feb 2015.

Click for updates To cite this article: Ü. Kizil, L. Genç, T. T. Genç, S. Rahman & M. L. Khaitsa (2015): E-nose identification of Salmonella enterica in poultry manure, British Poultry Science, DOI: 10.1080/00071668.2015.1014467 To link to this article: http://dx.doi.org/10.1080/00071668.2015.1014467

Disclaimer: This is a version of an unedited manuscript that has been accepted for publication. As a service to authors and researchers we are providing this version of the accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proof will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to this version also.

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

1

Publisher: Taylor & Francis & British Poultry Science Ltd Journal: British Poultry Science DOI: 10.1080/00071668.2015.1014467

CBPS-2014-345

us

cr

E-nose identification of Salmonella enterica in poultry manure

Department of Agricultural Structures and Irrigation, Agricultural Sensor and Remote

ed

1

M

an

Ü. KIZIL1, L. GENÇ1, T. T. GENÇ2, S. RAHMAN3 AND M. L. KHAITSA4

Sensing Laboratory, Çanakkale Onsekiz Mart University, Çanakkale, Turkey, 2 Department of

ce pt

Biology, Çanakkale Onsekiz Mart University, Çanakkale, Turkey, 3 Department of Agricultural and Biosystems Engineering, North Dakota State University, North Dakota, USA and 4 Department of Pathobiology and Population Medicine, Mississippi State University, Mississippi State, USA

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

Edited Hocking 28 January 2015

ip t

Ed. Kjaer, January2015;

Running title: E-NOSE IDENTIFICATION OF SALMONELLA

Correspondence to: Ü. Kizil, Department of Agricultural Structures and Irrigation, Agricultural Sensor and Remote Sensing Laboratory, Çanakkale Onsekiz Mart University, Çanakkale, Turkey. E-mail: [email protected]

Accepted for publication 20th December 2014

2 Abstract. 1. A DiagNose II electronic nose system was tested to evaluate the performance of such systems in the detection of the Salmonella enterica pathogen in poultry manure. 2. To build a database poultry manure samples were collected from seven broiler houses, samples were homogenised, and subdivided into four portions. One portion was left as is; the other three portions were artificially infected with S. enterica.

ip t

3. An Artificial Neural Network (ANN) model was developed and validated using the

cr

4. In order to test the performance of DiagNose II and the ANN model 16 manure samples

us

were collected from six different broiler houses and tested using these two systems.

5. The results showed that DiagNose II was able to classify manure samples correctly as

an

infected or non-infected based on the ANN model developed with a 94% level of accuracy. INTRODUCTION

M

Salmonella is one of the most important zoonotic pathogens that causes environmental and

ed

health problems in Europe (Danguy des De´serts et al., 2011) and worldwide (Hendriksen et al., 2011). More than 2200 Salmonella serotypes affect human health (Su and Li, 2004). Most

ce pt

of the Salmonella subspecies are mobile and produce hydrogen sulfide (Giannella, 1996) and they may cause typhoid, paratyphoid, and food poisoning (Ryan and Ray, 2004). The taxonomy of Salmonella is highly complex (Tindall, 2005). For instance, Salmonella enterica has 6 subspecies including enterica, salamae, arizonae, diarizonae, houtenae and indica. The majority of Salmonella isolated from humans (> 99.5%) is S. enterica (Herrera-Leon, 2007)

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

developed database.

which is commonly acquired from contaminated food and is an important cause of illness worldwide (Hendriksen et al., 2011). Zoonotic foodborne pathogens continue to burden the public health of the United States (US) and a major source of these pathogens remains animals and their waste products. These pathogens can enter meat and milk products during slaughter or at milking, or can contaminate raw vegetables when soil is fertilised with improperly composted (or

3 uncomposted) animal manure (McEwen and Fedorka-Cray, 2002). It is worth noting that Salmonella and Shiga toxin-producing E. coli (STEC) are some of the most important pathogens transmitted during manure disposal. Therefore, it is vitally important to rapidly detect the presence or absence of these pathogens before the disposal of manure to minimise environmental and public health concern.

ip t

It is possible to identify even a single cell by using conventional culture methods.

cr

analysis (Su et al., 2003). On the other hand, with the latest development in technology,

us

electronic nose systems are becoming widespread in different applications due to their fast and reliable results (Balasubramanian et al., 2005). Comparatively, the detection of these

an

pathogens using a standard microbial plate count is time consuming and expensive. The use of an electronic nose could provide a rapid and accurate means of identifying target bacteria

M

with little or no sample preparation. The metabolic activity of pathogens generates unique

ed

mixtures of by-products such as gases. These gases can be trapped in an electronic nose to identify the presence of pathogens in a matrix.

ce pt

There are various applications of electronic nose systems in agriculture and other disciplines. Ellis et al. (2004) used gas chromatography-mass spectrometry to identify bovine tuberculosis from cattle breath samples. These authors concluded that it is possible to differentiate between healthy and Mycobacterium bovis infected cattle from their breath derived volatiles. Kizil and Lindley (2009) developed a gas sensor array to estimate manure

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

However, these methods are time-consuming and may take up to a week to confirm the

nutrient contents. They were able to estimate manure N, P, and K contents with correlation coefficients of 0.80, 0.76, and 0.70, respectively. Additionally, Pavloua et al. (2002) used an electronic nose system to identify volatile production patterns of different bacteria in urine. Their study reported that use of electronic nose system has potential for early detection of urinary infections. In another study, D’Amico et al. (2010) was able to identify lung cancer from the breath samples of patients using an electronic nose system. Berna et al. (2013) used

4 a DiagNoseTM electronic nose system in order to detect and distinguish Escherichia coli O157:H7 and Salmonella spp. They were able to discriminate pathogenic Salmonella from E. coli O157:H7. Kim et al. (2010) used a surface acoustic wave (SAW) sensor based electronic nose system to identify Salmonella contamination on beef samples. They reported that Salmonella contamination could be detected after 4 h of incubation. In a similar study,

ip t

Balasubramanian et al. (2008) used an electronic nose system to predict Salmonella

cr

(PCA), stepwise linear regression, etc.

us

Currently there is little information available on the use of an electronic nose to identify presence of Salmonella in manure. Therefore, this study evaluated the suitability of

an

identifying Salmonella in manure using an electronic nose.

Gas sensors used in electronic nose systems generally are sensitive to a wide range of

M

volatile compounds. Thus, sensor responses create a partially overlapping sensitivity (Zaromb

ed

and Stellar, 1984). Therefore, a pattern recognition technique is generally required to identify volatile compounds (Gardner, 1991). Artificial Neural Networks (ANNs) are considered to be

ce pt

the most promising pattern recognition method and typically used in electronic nose applications (Gardner et al., 1992). Srivastava (2003) developed a tin oxide gas sensor array employed with an ANN model to detect and identify volatile organic compounds (VOCs) even in the noisy conditions.

Odour, air quality, and water quality are major environmental concerns from livestock

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

typhimurium with different statistical analysis methods such as principal components analysis

production facilities since they generate pollutant gases at different stages of production

systems and manure management. Similarly, manure borne pathogens may contaminate both surface and groundwater and affect both humans and livestock. Early detection of a disease increases the chances of disinfecting or adapting a better management practice that would not contribute to environment or any health risk. S. enterica is mainly observed in broiler and laying hen facilities. Manure management and housing systems play a significant role in the

5 spread of diseases caused by this pathogen; S. enterica is mainly transferred via manure disposal. If an electronic nose can be used to detect the presence of S. enterica early, it increases the chances of successful treatment. Thus, the objective of this study was to evaluate the performance of a commercially available electronic nose system in the detection of S.

MATERIALS AND METHODS

ip t

enterica in poultry manure.

cr

laboratory analysis protocols in 2011. Based on the promising results of that study, a research

us

project was developed at the Çanakkale Onsekiz Mart University (ÇOMÜ), Agricultural Sensor and Remote Sensing Laboratory with cooperation of, ÇOMÜ Department of Biology,

an

North Dakota State University, Department of Agricultural and Biosystems Engineering and, Mississippi State University Department of Pathobiology and Population Medicine.

M

Following methodology was followed in the research.

ed

Sample collection and microbiological analysis

Manure samples (mixture of manure, urine, rice hulls and other feed material) were collected

ce pt

from seven broiler houses located in Biga town of Çanakkale province, Turkey (these broiler houses use rice hulls as bedding material). The samples were collected in sterile bags and transferred to the laboratory within 2 h. They were coded as S1, S2, S3, S4, S5, S6 and S7, each representing a different facility, stored at 4 ºC and analysed within 24 h of collection. Each sample was divided into 4 portions of 43 g and labelled as B, C1, C2, and D. All

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

An initial study was conducted at the North Dakota State University to determine the

portions of samples were homogenised with equal amounts of Buffered Peptone Water (BPW). In order to determine the host S. enterica populations in the manure samples (0 h), 1 g of sample from B-portions was re-suspended in BPW, plated after serial dilutions and incubated at 37 ºC for 24 h. Both Mannitol Lysine Crystal Violet Brilliant Green Agar (MLCB) and Salmonella-Shigella-Agar (SS) were used for selective visualisation and primary counting of S. enterica species. The reminder of the B-portions of samples and sterilised D-

6 portions were incubated at 37 ºC for 48 h. C1 and C2 portions of manure samples were supplemented with 103 and 2×103 cells of S. enterica subsp. enterica, respectively, from the logarithmically growing culture and incubated at 37 ºC for 48 h. At different time intervals, 1 g samples were removed from all portions of manure samples, re-suspended in BPW and plated to nutrient agar for determining total bacterial population, and MLCB agar and SS agar

ip t

for determining S. enterica colony forming units (CFU).

cr

sample. Table 1 summarises the sample abbreviations and associated electronic nose readings Table 1 near here

us

and microbial analysis intervals.

S. enterica subsp. enterica (ATCC 13311) was obtained from Refik Saydam National

an

Type Culture Collection Laboratory. Control strain ATCC 13311 was plated on SS agar and incubated at 37 ºC for 18-24 h. For long term storage S. enterica was suspended in 20%

M

glycerol and stored at -80 oC. For determining logarithmic growth of S. enterica subsp.

ed

enterica, a single bacterial colony was taken from SS agar plate and inoculated to nutrient broth culture and incubated at 37 ºC for 10-12 h. Fresh nutrient broth culture was started from

ce pt

overnight culture and at 3 h time intervals one ml sample was taken for determining optical density (OD 630 nm) and to estimate bacterial numbers. Colony counts were done after serial dilutions and plating onto SS-agar. The optical density and the logarithmic of total viable cell count (CFU/ml) for S. enterica subspecies enterica population is given in Figure 1. A total of 84 electronic nose readings and corresponding laboratory analyses were

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

Electronic nose readings were conducted at the same time intervals for each manure

used to train and validate the ANN model that was used for sample classification. All samples were discarded after being autoclaved.

Figure 1 near here

Electronic nose system and sampling procedure In this study 43 g of collected manure was inoculated with S. enterica and placed in 100 ml sterile glass jars and covered with a lid to make it airtight. Jars were left at room temperature for 4-6 h after the inoculation to collect the headspace gas produced by the S. enterica. Three

7 replicates were carried out for both controlled and inoculated samples. Subsequently, as detailed in Table 1, headspace gas was collected and analysed using the DiagNose II ® (an electronic nose, the eNose Company, Zutphen, The Netherlands) as described below. Headspace gas sample analysis The DiagNose II system contains an array of 12 metal-oxide gas sensors. The sensor surface

ip t

adsorbs the oxygen causing an oxidation that result in lower levels of conductivity. As the

cr

sensor changes as well. The sensor material, sensor temperature dynamics and the chemical

us

reaction rates effects the conductivity change. The conductivity changes during a certain time is recorded, and then used as the signal for a particular sample. The software enables

an

downloading the data in CSV format for further statistical analysis.

The system is purged with either ambient air or air that has been led through an air

M

filter. In this study a standard syringe filter was used to prevent particles and/or liquids

ed

entering the system. As the system is turned-on, it purges sensors with filtered ambient air for about 120 s. After the initial purge, the headspace volatiles are re-circulated for 120 s (pump-

ce pt

on). After re-circulation, sensor responses are recorded for another 120 s with pump off. At last the system is re-purged for 120 s in order to clean any residual effect from the previous sampling event. Thus, one sampling starts with activation of the pump and ends with a system purge (Figure 2).

Figure 2 near here

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

amount of oxygen decreases due to reaction with other substances the conductivity of the

Data processing and database building DiagNose II employs 12 gas sensors; however only seven sensors responded to the

manure samples. In order to check sensor response, the system was tested with different materials including perfume and vinegar and all sensors were performing well. In a similar problem Balasubramanian et al. (2005) reported that some gas sensors are sensitive to polar compounds such as water vapour thus they may not respond to such compounds. Also, sensitivity of these sensors to aromatic and halogenated hydrocarbons is not good enough (Ho

8 et al., 2001). The operation temperature and composition of the metal oxide active material for each sensor is different. Therefore, each sensor provides different responses to various gases (Bartlett and Gardner, 1992, and Ho et al., 2001). Hence, it is highly possible that some sensors did not respond to manure headspace gases. For this reason, data from the 7 sensors only were considered in the database.

ip t

Following the sample readings all data were downloaded to a computer in CSV format

cr

processing. As explained above there were 84 readings and each belonged to a sample

us

containing different S. enterica populations. Each reading also included 7 sensor responses to provide a total of 588 sensor responses.

an

The e-nose system uses fixed measurement intervals. As standard modulation scheme a sinusoidal period of 20 s is used. The database consists of responses of 588 gas sensors.

M

None of the sensor responses showed considerable signal noises. Therefore, the smoothing

ed

method does not affect the results considerably. Since the signals do not require major smoothing a simple moving average method was employed with a minimal value of M (M=2)

ce pt

in the equation:

y[i] =



x[i + j]

(1)

Where [xi] is a sensor response at s (second) intervals i (from 0 to 360); M is the

number of readings that were used in the averaging and j is the number of the last reading.

Ac

Downloaded by [University of Sussex Library] at 13:54 09 February 2015

for data processing and classification. The data were then saved as XLS files for further

The signals were then normalized using equation 2.

=

(2)

Where Vn is the normalized sensor response at the ith s; Vi is smoothed sensor response recorded at s i and Vmin is the minimum sensor response recorded during the reading (Balasubramanian et al., 2005). An exsample of raw, smoothed, and normalized sensor response curves is presented in Figure 3.

Figure 3 near here

9

Next, areas of curves that belong to sensor responses were calculated to create a numeric database. In the area calculation, a simple sum method was employed. In this approach, the total area under a curve is the sum of sensor responses recorded each s. After converting all sensor responses to a numeric database, two matrices were developed including sensor response (S) and corresponding laboratory analysis (A).

,

,

=

. . . . .

ip t

,

(4)

,

an

In the A matrix laboratory analysis results were not provided as numeric values, but as a vector of infectivity (“infected” or “uninfected”). Initially all the manure samples were

M

analysed to determine the S. enterica populations without any treatment. Laboratory analysis showed that manure samples contained S. enterica populations between 3.52×104 and

ed

3.53×106. In the A matrix, manure samples with microbial counts >106 were considered as “infected”, samples with microbial counts

E-nose identification of Salmonella enterica in poultry manure.

A DiagNose II electronic nose (e-nose) system was tested to evaluate the performance of such systems in the detection of the Salmonella enterica patho...
710KB Sizes 0 Downloads 7 Views