Morris et al. [25] found that of all the samples taken from a commercial broiler operation, feed samples were most frequently contaminated with Salmonella. Human outbreaks of Salmonellosis have been traced back to feed for decades. In 1958, an outbreak of infection of S. Hadar in Israel was linked to the consumption of chicken liver and was eventually traced back to bone meal fed to the chickens [26]. Frozen chickens from a packing plant in Cheshire, England, were implicated in a large outbreak in 1968 of infection with S. Virchow [27]. The investigation showed that the hatchery and some rearing farms that supplied the packing plant contained chickens colonized with S. Virchow. In this investigation, the same serotype of Salmonella was isolated from feed fed to the chickens [28]. Chickens served in a restaurant in Arkansas caused an outbreak of S. Agona. The chickens were traced to a farm in Mississippi that fed the chickens with feed containing Peruvian fish meal found to be contaminated with S. Agona [29] The fish meal was found to be the ultimate source for a number of S. Agona infections in the United States, the United Kingdom, Israel, and the Netherlands.

Several more recent investigations using sophisticated genotyping methods have found confirmed that Salmonella in feed can be a primary source of contamination. In a study by Shirota et al. [30], S. Enteritidis strains obtained from feed samples and egg contents taken from a layer farm showed pulsed-field gel electrophoresis (PFGE) patterns that were genetically related. Futhermore, the isolates belonged to a single phage type which suggested that the contamination of the farms was linked to the occurrence of salmonellae in feed. Using PFGE, Wasyl et al. [31] found Salmonella isolates with identical pulse-types isolated from feed and poultry. Bucher et al. [32] also used PFGE along with serotyping, phage typing and antimicrobial resistance typing and concluded that Salmonella strains isolated from broiler feed were indistinguishable from strains isolated in packaged raw, frozen chicken nuggets and strips.

Feed is typically comprised of corn, soybeans, oats, alfalfa, calcium and a vitamin mixture [33]. This composition may vary depending on the manufacturer and the type of poultry being fed. For example, laying hens require higher concentrations of calcium for egg shell production. To produce the feed, ingredients are mixed and steam processed. After processing, the feed may be cooled by passing through a cooling air unit or, prior to cooling, pelleted into a cylinder like shape.

Himathongkham et al. [34] demonstrated that feed moisture and conditioning time were two factors that play a crucial role in the lethality of the pelleting process for bacteria. Most studies agree that the pelleting process is more effective at reducing Salmonella contamination. Cox et al. [24] reported that 92% of mash feed samples were positive for Salmonella but no pelleted samples were positive. However, Veldman et al. [18] found 21% of mash feeds and 1.4% of pelleted feeds were positive for Salmonella. Similarly, Jones et al. [19] found that 8.8% of mash feed samples and 4.2% of pelleted feed samples were contaminated with Salmonella.

If Salmonella is destroyed during the heat treatments, the possibility of re-contamination still exists. Raw feed ingredients can serve as a source of contamination to the plant environment and ultimately to the final feed product. Veldman et al [18] sampled raw feed ingredients and found 130 samples of fish meal (31%), 83 samples of meat and bone meal (4%), 58 samples of tapioca (2%) and 15 samples of maize grits (27%) were positive for Salmonella. The data presented by Jones et al. [19] indicated that dust within feed manufacturing facilities could serve as a major source of contamination to the final product. The authors suggested that mechanical vibrations and air currents around the pellet mill might have resulted in dust particles being dislodged and landing on the final pelleted feed. Davies and Wray [22] showed that the cooling unit was colonized by Salmonella which might serve as an airborne source of contamination to the final feed product. In addition, there is the possibility of feed being contaminated during transportation and/or storage [35]. With the possibilities of post-process contamination, detection methods are critical for preventing flock contamination. The next section will address current detection methods available and possibilities for future developments.

Detection of Salmonella in feed can be challenging due to low number of cells present in a large volume of feed. Riley [36] estimated a contamination rate of feed passing through a contaminated cooler would pick up 1 Salmonella organism per 10 to 100 tons if the facility was not receiving feed ingredient loads that were contaminated. At such a level of contamination, the challenge becomes designing both a sampling program and a method of detection that can detect 1 cell in 10 tons of feed.

Complications of isolating Salmonella from feed not only has been suggested to stem from the non-uniform distribution of the organism within the samples, but also from the effect of stress on the organisms from processing treatments used in feed mills [18,19,37]. In addition, the treatment of feed with formaldehyde can interfere with detection methods and give a false negative result [38].

Numerous detection methods have been developed for Salmonella such as culturing, immunological methods and nucleic acid based methods [38]. Typically, the method is chosen based on the application of the user. For example, if the desire is to not only detect but also to characterize Salmonella, the isolate will need to be recovered by culturing for further genotyping, antibiotic resistance typing, serotyping or other characterizations. However, if only presence or absence is necessary than nucleic acid detection assays are sufficient. Each method has advantages and disadvantages and every method has some limitations. The following sections will discuss the methods available and describe shortcomings and benefits for using a particular assay when applied to poultry breeder feeds.

Traditional microbiological culture methods for the detection of Salmonella in feeds include selective enrichment and selective and/or differential plating. Culturing methods for the detection of Salmonella have been reviewed [39,40]. In general, plate agar media contains a pH indicator and lactose to differentiate Salmonella (a fermentor) from non-fermenting bacteria. Since most Salmonellae are hydrogen sulfide producers, tergetol can be added which results in Samonella colonies turning black. Other selective agents may be added such as antibiotics like novobicin which permit the growth of Salmonella while inhibiting competing microorganisms.

Chromogenic media work by using enzyme substrates that release colored dyes after hydrolysis, resulting in Salmonella colonies being colored and easily differentiated from other flora. This type of agar has an increased specificity over conventional media. However, some reports have shown that conventional media are less inhibitory and therefore more sensitive because stressed or injured microorganisms can be recovered [41,42]. Regardless of the plating media used, culturing methods are often considered as the “gold standards” but are time consuming in that they require days for results. Given that infection of an entire flock can occur in as few as three days [43], a method of detection that is more rapid than culturing is needed to implement control measures and control any further spread of infection.

The genus of Salmonella consists of only two species, S. enterica and S. bongori but over 2,500 serotypes [44]. There are 47 possible O-antigens (lipopolysaccharide of cell wall) and 60 possible H-antigens (flagellum) with each serovar having its own unique combination of O- and H-antigens [45]. The name of each serovar was given based on the syndrome displayed, host specificity or geographical location [45].

Each serovar of Salmonella may vary widely in characteristics including severity of disease, virulence properties, ability to colonize chickens and survival in the environment outside the host [46–49]. For this reason, identification beyond the species level is necessary. Furthermore, serovar information is used by local and state health departments and CDC to monitor local, regional, and national trends of salmonellosis.

Serological based assays (immunoassays) such as ELISA have been widely used for Salmonella detection because they allow the sensitive and specific detection [50]. Immunological methods for the detection of Salmonella in animal feeds have been reviewed [40]. Immunoassays offer the ability to detect and distinguish serovars of Salmonella. However, immunoassays are hard to incorporate into an array format for multiple targets because high level of cross-reactivity between antibodies limits the number of targets that can be detected on the same array. Use of immunomagnetic beads (IMBs) in detection assays has been reported, and it shows great potential in Salmonella detection from complex matrices or environmental samples [51,52]. In IMB applications, however the number of target analytes detected in a single assay has been restricted to one or two due to high cross-reactivity and limited availability of commercial IMBs or antibodies for specific target organisms.

Polymerase Chain Reaction (PCR) has been gaining popularity as a tool in microbiological diagnosis due to the efficient, rapid and sensitive methods of detection. The methodology of PCR for the detection of foodborne pathogens has been reviewed previously [53]. Several variations of standard PCR, such as multiplex PCR and real-time PCR, have recently been employed for Salmonella detection, and these methods have provided high sensitivity with some assays being able to detect as few as 30 cells per sample [54]. The important criteria in the development of a nucleic acid based detection assay for Salmonella is the ability to detect all the diverse serotypes of the organism and PCR has been employed to replace conventional serotyping methods [55]. PCR-based serotypings depend on specific virulence genes, and have provided high specificity [56,57]. However there is a limitation on the number of target Salmonella serovars which can be detected in single PCR reaction. Even in multiplex PCR, it is difficult to incorporate more than five to six primer sets (correlating to five or six serovars) in one reaction due to cross-reactivity. Considering that there are at least 12 serotypes of Salmonella commonly associated with poultry [58], there is a clear need for an assay able to simultaneously detect multiple Salmonella serovars with minimal cross-reactivity.

Biosensors are being developed because they offer rapid and sensitive unconventional detection methods. Biosensors generally contain two components, a biological material (nucleic acid or antibody) closely associated with a transducing system. The transducer emits a signal when the target is captured that can be optical, electrochemical, thermometric, piezoelectric, magnetic or micromechanical. Biosensors differ from conventional detection methods because they are self-contained single units that have both the detection and reporting components.

Optical biosensors which utilize a fluorescent signal are typically the most common type of sensor [59]. However, biosensors using transducers other than optics have been developed for the specific detection of Salmonella. Olsen et al. [60] utilized bacteriophage specific for Salmonella typhimurium. In this biosensor, the capture of bacteria by bacteriophage that was adsorbed to a piezoelectric transducer resulted in a resonance frequency change measured with a Maxtek acoustical wave device. Su et al. [61] used an antibody bound to a gold coated quartz crystal surface with a gold electrode as a biosensor. After capturing Salmonella, changes in high-frequency impedance were directly correlated to the number of captured Salmonella cells. Pathirana et al. [62] and Kim et al. [63] also used a similar impedance analysis to create biosensors for the detection of Salmonella typhimurium.

DNA microarrays are a type of optical biosensors that detect hybridization of DNA sequences between bound probes having known sequences to fluorescently labeled DNA from an analyte. If hybridization is successful, a fluorescence signal is emitted upon excitation with a laser and the intensity of the signal can be used for quantification and identification of the analyte. Planar DNA microarrays allow thousands of specific DNA sequences to be screened simultaneously on a small single glass slide. Using DNA microarrays, multiple Salmonella serovars can be concurrently detected and the presence of virulence genes and antibiotic resistance genes can also be identified at the same time [64,65]. While planar microarrays offer the great potential for a rapid and sensitive detection of multiple pathogens [66,67] high fabrication cost and requirement for expensive equipments have been limiting their wide application in routine applications [68].

Bead-based microarrays, as an alternative to planar microarrays, have been developed to perform multiplexed detection assays [68–70]. In bead-based arrays, microspheres are employed as solid support for the capture molecules (e.g., antibodies, oligonucleotide probes), instead of glass slides conventionally used in planar microarrays (Figure 2).

The individual microspheres are color-coded by distinct fluorescent dyes. In each DNA microarray, the oligonucleotide probe is immobilized to the surface of a distinct type of microspheres which are chemically functionalized. Different bead sets are then pooled to create a library, and hybridization is performed in a single vial or single well in a 96-well microplate containing the library of all bead types. After hybridization, presence of targets can be detected with a two-laser flow cytometer, where one laser interrogates the encoding dyes of beads to determine the probe identity and the other laser determines the presence of targets in the sample by reading the second fluorescent signal from hybridized targets.

Bead-based DNA arrays have several advantages over planar microarrays; (1) they can accommodate standard 96-well sample preparation systems; (2) since probes are coupled to distinct microspheres, each hybridization reaction can be analyzed; (3) if an additional target has to be included into the assay, a new type of probe-loaded bead can simply be added to the array unlike planar microarrays which require the new fabrication of arrays to add a target [68,69]. Bead-based arrays coupled with flow cytometry technology have been successfully applied for the simultaneous detection of multiple bacterial pathogens [71]; however this study was done with pure culture of target pathogens. Bead-based arrays have never been employed for detection of pathogens in more complex matrices such as feed or environmental samples. Bead-based arrays have been more commonly used in clinical applications such as simultaneous quantification of cytokines or autoantibodies from biological samples [72–75]. Bead-based arrays have the great potential for rapid and sensitive identification of Salmonella from feed. The criteria for optimal bead based array design are listed in Table 1.

An initial step for rapid microarray development involves the selection of target genes and design of probes and primers that detect and characterize Salmonella spp. which are commonly found in poultry breeder feeds. However, effective sample preparation methods to minimize the effect of environmental factors are usually required to retrieve Salmonella from feed matrices. Immunomagnetic separation using anti-Salmonella magnetic beads can be employed as a standard method to separate Salmonella from feed matrices [76]. Cultural pre-enrichment also can be utilized to optimize sample preparation to alleviate any inhibitory effect from feed matrices while keeping the total assay time short.

Both optical and electrochemical biosensors offer advantages, but also come with disadvantages. Optical techniques have been demonstrated to provide better sensitivity than electrochemical ones [59]. Electrochemical techniques offer simplicity over optical detection methods. However, optical techniques offer the ability to capture and detect many targets and for this reason are usually more costly. Some biosensors are sensitive but they still are not capable of the same detection levels as traditional techniques.