Poultry industry paradigms: connecting the dots

From Mendelism to Genomics

Initial selection programs for poultry were based on simple Mendelian principles, using phenotypic mass selection; later improved with the advent of statistical methodologies such as the Selection Index (Hazel, 1943) and followed by the implementation of the mixed model equations and the animal model (Henderson, 1984; Quaas and Pollak, 1980; Searle, 1991). These were very efficient tools for the prediction of breeding values used in selection, in the form of the best linear unbiased predictor (BLUP), modernizing the genetic improvement of poultry populations.

Genomics has evolved from a small number of Restricted Fragment Length Polymorphisms and Microsatellites and low/medium-density Single Nucleotide Polymorphism (SNP) chips to whole genome sequence with tens of millions of data points (Wolc, 2022). The release of the chicken genome (International Chicken Genome Sequencing Consortium, 2004) allowed the development of molecular tools and the implementation of marker assisted selection (MAS) and genomic selection (GS) programs (Dekkers, 2004; Meuwissen et al., 2001). An informative summary of genotyping platforms and implementation of genomics by the poultry industry was presented by Wolc et al. (2016).

This was further improved by combining pedigree and genomic information of individuals into a joined relationship matrix and single-step prediction method (Legarra et al., 2009; Misztal et al., 2009) as the foundation for the establishment of a new paradigm of genetic improvement of pure lines used in the production of commercial broilers and layers.

Phenotype is King

One of the founders of livestock breeding programs (Dr. Jay Lush) used to say that data was the raw material used by the geneticists to improve farm animal populations. Genetic improvement is possible thanks to phenotypes recorded as quantitative or categorical values, compiled in datasets, to feed into software for predicting breeding values. Contrary to the original expectations, practical implementation of genomic selection requires the continual collection of good quality genotypes and phenotypes and model retraining to maximize its benefits (Weng et al., 2016). Breeding goals evolved from simply recording body weight for broilers, and counting eggs for layers and utilizing family means to more sophisticated evaluation programs for broilers and layers, including different components of the production (PD) curve, efficiency (feed intake and feed conversion), product quality, fitness, and reproduction, behavioral and disease-related traits. More recently, breeding goals have expanded to include welfare related traits such as bone strength, gait, which are measured using novel technologies, such as x-ray, CT-scan, iStat, automated camera gait scores, hypobaric chambers, and specific disease trials (Wolc, 2022); as well as behavior-observation traits, such as temperament, flightiness, and perching and nesting behavior.

Moreover, data collection must adapt to changing housing systems to collect individual data in group housing: feeding stations for broilers, automatic nests for layers, and sensors to track behavior, among others. The importance of sustainability for food security is reflected in the evolution of breeding goals. Neeteson-van Nieuwenhoven et al. (2013) presented a graphic representation of the relative importance of productivity, feed efficiency, environmental efficiency, human health, product quality, adaptability, robustness, and reproduction for a 75-year horizon (Figure 1). The graph offers an interesting perspective from the evolution of breeding goals, which mainly emphasized productivity (1950), to complex and balanced set that include all the aspects mentioned above (2015 to 2025). In the future, we may rely on advance technologies, such as vision-devices combined with artificial intelligence to process big data for real-time collection. These advancements would enable tracking of bird movements, navigation of housing systems, utilization of nests, feeding and drinking spaces, and predict or notify about potential health issues. These large data collections in combination with genomics, for instance, allows to create predictive models for different traits and events. Long et al. (2007) used machine learning to identify 17 SNPs associated with chick mortality in broilers with success. In other species, machine learning and the use of artificial neural network methods was shown to be effective in the prediction of disease phenotypes, estimation of genomic breeding values for disease susceptibility, prediction of production, feed intake, mortality rates, and fertility (Nayeri et al., 2019).

1. Download : Download high-res image (167KB)
2. Download : Download full-size image

Figure 1. Adapted from Neeteson-van Nieuwenhoven et al., 2013. Radar plots showing the relative importance of its elements based on the distance from the origin for each polygon angle; elements on opposite axes are not implied to be antagonistic.

Selection Priorities Then and Now

The industry started with dual-purpose birds, which were grown for meat and produced eggs in sufficient quantities to satisfy a limited demand. Later the demand for poultry products increased dramatically worldwide, as illustrated earlier, and the industry-focused on maximizing production efficiency and implementing mass production of poultry products to increase availability and drive down costs for the consumer. This evolution has opened the poultry product markets to diversification and sophisticated stratification into specific niches, and genetic selection played an important role in it. For example, the subjective evaluation of eggshell color and consumer preference has led to a selection for this trait, even though it is not related to an increase in nutritional value (Cavero et al., 2012; Rondoni et al., 2020).

The diversification of poultry product portfolios and the requirements of incorporating animal welfare and behavior-related traits to the breeding programs of layer and broilers have transformed the selection goals. In broilers, these goals went from primarily grow, carcass traits, and efficiency to a complex system that must include production, reproduction (breeders’ fertility and hatchability), product quality, and traits related to animal welfare and environmental impact. Similarly, in layers, there was a shift from a focus on egg production, egg quality, and efficiency traits to programs that also include robustness, disease resistance, behavior, and indicators of welfare, such as osteoporosis, plumage condition, among others. This trend is expected to continue as new technologies become available to identify better healthy and robust birds, those adaptable to a wide variety of production environments with efficient utilization of resources.

What Have We Achieved So Far and Where Do We Go From Here?

For egg-type birds, good summaries of genetic achievements were illustrated by Flock (2009), Arthur (2015), and Preisinger (2018); the evolution of 53 years and 37 editions of the random sample tests in North Carolina was summarized by Anderson et al. (2013); while, 50 years of environmental impact (1960-2010) was presented by Pelletier et al. (2014) in the US. Published data have shown continuing improvements in egg production (greater than 30% increase in the rate of lay and over one month decrease in age to 50% PD), reduction in body weight (up to 60% decrease),and feed consumption. Over the same time, there was also increase in egg weight (up to 1 g per egg), feed conversion improvement (up to 50%), an increased livability (up to approximately 10%), and improved egg quality for commercial varieties of both white and brown eggs. Pelletier et al. (2014) documented the lowered environmental impacts of the modern layers by reduction of inputs (26% and 32% lower feed and water consumption, respectively) while reducing the environmental footprint by lowering the acidifying, eutrophying, and green-house gas emissions by 65, 71, and 71%, respectively; and with a lower accumulated energy demand of 31% than the 1958 to 1960 layers. All these changes have continued throughout the 50-plus year data and they have been driven by genetic gain.

In broilers, Havenstein et al. (2003) compared the performance of 1957 (ACRBC) vs. 2001 (Ross) lines fed a typical 1957 diet to prove that a large proportion of progress in broiler performance was due to genetic improvement. This publication documented a difference of more than 3-fold in 8-wk body weight and more than 2-fold in carcass weight in favor of modern broilers, which also showed advantages in carcass and breast yield of 13.6 and 9.9 points percent. A long-term divergent selection experiment for low or high 8-wk body weight in chicken lines at Virginia Tech (initiated by Dr. Paul Siegel) illustrated, quite elegantly, the impacts of genetic response to selection for the initial 27 generations (Dunnington and Siegel 1985). These lines have been under divergent selection for about 60 generations and have been the source of multiple scientific publications showing different aspects of this long-term selection experiment. Hill et al. (2016) reviewed the achievements in poultry breeding (layers and broilers) for sustainability by maintaining and enhancing multi-trait genetic improvement. They concluded that sufficient genetic variation, required for continued response in production traits and traits related to health and welfare, has not declined in pure line populations.

The successful story of genetic improvement in poultry has no comparison to other livestock species, and its future is bright. The large breeding companies, for broilers and layers, are implementing sophisticated genetic breeding programs. Unlike in the past, the current breeding programs balance productivity and fitness traits and keep in perspective consumer demands, public perception, and the need to ensure sustainable and welfare-friendly production systems while controlling the cost of production of poultry protein for the increasing human population. In order to reach these complex societal requests, it is important to continue to implement data collection of animal welfare-associated phenotypes and genomic features combined with advanced analytical methods for the best accuracy of predicted breeding values. The use of advanced molecular and reproductive techniques are possible but contingent on consumer acceptance and governmental regulations.

Data Sciences for Addressing Poultry Production Issues

As previously mentioned, animal welfare is an important attribute of the food quality concept. Consumers expect their animal-related products to be produced concerning to the welfare of the animals (Butterworth et al., 2009; Swanson, 2010; Daigle et al., 2014). In a commercial broiler grow-out house, up to ten thousand birds grow in a controlled environment. Living space/density, food/water supply, indoor thermal environment, litter/air quality, lighting, and group activity are affecting their welfare, health, and growth (Dawkins et al., 2004; Bessei, 2006; Brown et al., 2008; Mench, 2008; Buijs et al., 2009; Hale et al., 2010). The broiler industry (e.g., producers, retailers, and restaurants in the US and EU) care about animal welfare in each stage of broiler production, from hatchery to grow-out and processing. For instance, National Chicken Council recommends that all member producers or processors follow the broiler welfare guidelines for assuring the humane treatment of animals and to promote the production of quality products (NCC, 2017). However, the daily routine task of poultry welfare inspection has proceeded manually in the commercial poultry facility at present, e.g., the “6-point Kestin system” suggested by Kestin et al. (1992) and the “3-point scoring system” used in the US (Webster et al., 2008), which is labor-intensive and time-consuming.

Sensing technologies, such as ultra-wideband, radio frequency identification (RFID), accelerometer, and computer vision-based monitoring, have been tested for livestock and poultry farming system (Dawson et al., 2007; Quwaider et al., 2010, Wang et al., 2019). Many monitoring methods require direct contact with birds (e.g., RFID and accelerometers) that potentially will affect animal activity/behavior and welfare. Thus, non-invasive/non-contact digital image analysis methods (i.e., cameras and automatic image processing) is considered an acceptable approach (Borgonovo, 2009). The computer vision-based (phenotyping) technology for monitoring behaviors of large animals (e.g., cattle and pigs) has been well developed (Nakarmi et al., 2014; Poursaberi et al., 2010; Viazzi et al., 2014). However, it is technically challenging to monitor smaller size individual animals such as broilers in commercial houses (e.g., 25,000-30,000 birds per house).

Currently, vision-based monitoring studies are focusing on single activity or behavior of broilers or layers such as feeding and drinking (Li et al., 2017), light preference (Riber, 2015; Liu et al., 2017), perching (Wang et al., 2019), pecking (Hartcher et al., 2016), dust bathing (Campbell et al., 2016), and group activity/response to water sprinkling cooling (Chai et al., 2018c, 2019b). Among previous studies, two groups distinguished out in system design and test. Dawkins et al. (2009, 2017) developed an “optical flow” method for measuring broiler welfare based on optical flow statistics of flock movements recorded on video. Aydin et al. (2010) integrated a monitoring system, “eYeNamic” for gait score monitoring in the broiler house. The basic principle of “optical flow” and “eYeNamic” is 2D imaging analysis of photo’s pixel change over time. Those early studies provided a blueprint that the future broiler welfare evaluation can be conducted with computer or machine vision-based methods. However, no existing system/method can track, identify, and locate individual birds with welfare concerns in real-time so far. Besides, previous studies (i.e., 2D imaging) collected information is insufficient to accurately quantify the welfare indicator (e.g., gait score). 2D imaging results need verification because broiler welfare indicators such as gait score are affected by multi factors (e.g., leg issues, infections, and body weight) (Dawkins et al. 2017). In addition to gait score, there are many other health or behavior-based welfare indicators (e.g., footpad lesion, broken leg/wing, hock or breast burn, breast blister, ascites, and dust bathing/foraging) that need to be monitored automatically (Shepherd and Fairchild, 2010). Therefore, advanced animal phenotyping technologies (e.g., an integrated 2D, 3D, thermal, and hyperspectral imaging system) are needed to capture more information on broiler health and behaviors to improve the machine vision-based broiler welfare evaluation.

Future Technologies for Poultry Welfare Management

To address issues related to poultry welfare management, a new generation of data management and scanning system that periodically moves above the broiler living space to evaluate the real-time individual and group broiler welfare and health status in the commercial grow-out house is required. The system should be able to track, identify, and locate individual birds with welfare concerns. Advanced phenotyping technologies (i.e., 2D, 3D, thermal, and hyper special imaging) will be applied to collect image information on animal health and natural/social behaviors. Image information can be transferred and stored in an on-site computer or cloud for cloud computing. Deep learning-based algorithms in on-site computers or edge computers through cloud systems can be used to analyze collected animal health and behavior information for quantifying key welfare indicators (e.g., gait score and footpad lesion) immediately and accurately. This kind of real-time poultry welfare evaluation tool can help with early detection of abnormal individual/group broilers, so caretakers/producers can take measures to protect animal welfare and health immediately. The scanning system is expected to cause minimal impact on broiler flock and house management.

For data analysis, a convolutional neural network (CNN) is one of the most applied deep learning algorithms for monitoring behaviors of chickens, such as feeding, drinking, and floor distribution (Guo et al., 2020, 2021; Li et al., 2021). Currently, researchers at the University of Georgia are using “you only look once” (YOLO) serials of CNN models in monitoring behaviors of broilers (Figure 2) and cage-free layers. Those automatic data management strategies should be further innovated for developing the core component of an integrated imaging system for evaluating poultry production and wellbeing in commercial broiler and cage-free layer houses.

1. Download : Download high-res image (1MB)
2. Download : Download full-size image

Figure 2. Broiler floor distribution monitoring.

Sustainability, Precision Feeding, and Nutrition

In feed formulation, improved feed efficiency and low cost per kilogram of carcass meat yield are the major goals for commercial nutritionists. These factors drive all aspects of sustainability, as nutritionists start formulating to more precisely meet the bird’s daily nutrient requirements, resulting in the production of more meat with less inputs, and reducing the carbon footprint and cost of feed. Furthermore, precision nutrition is a way to improve animal welfare, which is an important pillar of sustainability, as previously mentioned. In order to achieve this level of precision, it is important to continuously improve the methods for evaluating the nutrient value of raw ingredients, invest in the research of different feed additives, as well as determine the animal requirements of the constantly evolving genetic lines.

With that in mind, it is expected that the formulation of diets will shift from a metabolizable energy (ME) based-system to a net energy (NE) system-based. Net energy has been successfully used in swine nutrition as it allows for a more accurate prediction of production performance. However, currently, NE has a smaller advantage over ME for poultry, which is partially attributed to the lower dietary crude fiber content and hindgut fermentation in poultry compared to swine (van der Klis and Jansman, 2019). With the continuous improvement of prediction equations and digestibility assessments, considering sex, age, and genetics, it is possible that NE will become a precise tool to be routinely utilized by poultry nutritionists.

It is also expected that diets will be formulated with more synthetic amino acids beyond the currently economically available methionine, lysine, threonine, valine, arginine, and tryptophan that are selectively in use today. This could reduce, to some degree, the crude protein level in the diets, helping to reduce feed costs. Additionally, the reduction in nitrogen excretion with lower crude protein diets is important from an welfare standpoint (Shepherd and Fairchild, 2010). A study evaluating the incidence and severity of pododermatitis in broilers showed that high protein diets and all-vegetable diets caused footpad lesions that were 2 and 3 times more severe than low-protein and diets with animal by-products, respectively (Nagaraj et al., 2007). On an environmental standpoint, however, more research is needed. Benavides et al. (2020) performed a life-cycle analysis (LCA) evaluating the GHG emissions from diets containing different ingredients (corn, soybean meal, distiller-dried grains with solubles; DDGS, and synthetic amino acids). The LCA has shown that, individually, the amino acids methionine, lysine, and threonine had the highest carbon intensity per kg of ingredient produced (gCO_2e/kg). According to the authors, this is because of the material inputs and their intensive production process, which demands high energy. Despite showing the greatest carbon intensity amongst the main ingredients in these diets, when considering the complete feed, DDGS was they major contributor to the GHG emissions (Benavides et al., 2020).

The usage of feed additives, such as enzymes, will continue to be refined and improved, allowing better digestion of nutrients, and computer modeling programs will be utilized more in the future to help the poultry nutritionists to optimize feed formulation by focusing on profitability. Moreover, as birds are grown to larger market weights for deboning or roaster markets, separate sex feeding will become more important. This will allow the nutritionist to more precisely meet the nutrient requirements of each sex, avoiding waste of nutrients and resources. Despite the great achievements from the past decades, more research evaluating the relationship between nutrition and sustainability is needed. This area is rapidly evolving as companies place greater emphasis on environmental constraints in feed formulation and it will have a direct impact on feed procurement decisions and nutritionist options in the future.

Ingredients and feed analysis

It is highly likely that, in the future, diets will continue to include the current list of typical ingredients such as corn, wheat, grain sorghum, barley, soybean meal, canola meal, animal protein meals, distiller`s dried grain with solubles (DDGS), bakery byproduct, fats and oils, among others. Even more so, considering that a greater focus has been placed in improving the production of these ingredients in a sustainable way. For instance, great development has been seen in the soybean and corn production systems, with a reduction in the GHG emission and energy use by 2.6 and 4.5 kg of CO₂ per bushel and 9.6 and 13.3 kwh per bushel, respectively, from 1980 to 2020 (Field to Market, 2021). It is also possible that the crop genetic companies will improve the nutrient content, digestibility, and be able to maintain or increase the yield of the current crops. This will allow farmers to produce more nutrient-dense ingredients on less land and with less inputs, such as water and fertilizer. These new improved grains and oilseeds will need to preserve their identity to capture the nutrient and cost value. The ethanol industry has done ongoing work to produce DDGS with 30% or more crude protein and less fiber than the standard DDGS. Moreover, some novel ingredients, such as insect meal, algae, and seaweed, could be approved by the Food and Drug Administration (FDA) and become cost effective in the future.

Laboratory analysis of incoming ingredients and outgoing feed is critical today, and it is expected to be even more important in the future. The nutritionist needs an accurate analysis of each feed ingredient to characterize it correctly in the formulation matrix. The routine use of wet chemistry, near-infra-red (NIR), and other technologies will be imperative to help adjust the formulation quickly based on the actual feed ingredient quality. With that in mind, as new ingredients are being introduced to the diets, it is important to have a large dataset that is being updated periodically to improve calibration for amino acids, fatty acids, and other key nutritional parameters. Additionally, finished feeds will be required to be analyzed at the mill to confirm it meets the specifications for the birds it is intended to feed.

Formulations for Niches Markets

It is expected that niche markets, such as “no antibiotic ever” (NAE), “all vegetable diets”, slow-growing programs, GMO-free diets, organic, free-range, and diets for designer meats (e.g., extra omega fat content) will continue to exist, considering that they are often consumer-driven. While these marketing programs are in place today, the nutritionist will be continuously challenged to find the ingredients to support the future growth of these businesses, maintaining profitability and sustainability compared to the conventional programs. Recent studies have been focusing on the nutraceutical properties feed ingredients, such as amino acids, vitamins, minerals, and fatty acids in improving health (Lee et al., 2019; Castro and Kim, 2020; Alagawany et al., 2021; Kim et al., 2022). However, more research is needed to understand how current feed ingredients can contribute to animal health, especially considering systems in which antibiotics cannot be used. Additionally, there will be an increased need for the accreditation of these niche programs by the USDA or a third-party entity, ensuring their quality.

Feed Additives and Regulation

The use of different feed additives such as probiotics, prebiotics (e.g., mannan-oligosaccharides and beta-glucans), postbiotics, phytogenic compounds (e.g., oregano oil and saponins), trace elements, and short and medium chain fatty acids to help with gut health and improve live performance will continue to be refined. While these products cannot replace antibiotics, they can help the birds to cope with different stressors as well as assist with their recovery. Because of the large amount of available feed additives and their combinations in the market, nutritionists often struggle when deciding on what to include in their formulations. Furthermore, it is not rare that the results observed in the field are somewhat inconsistent with experimental set-ups, as they fail to show a benefit and some products are not cost-effective. Additionally, government agencies such as the FDA and USDA will need to work with the feed industry to approve new advances made in nutrition when they are proven safe and efficacious. Thus, nutritionists will need to continue to evaluate different product options that fit their nutritional goals, are safe, and are economically feasible, as there is an increasing need to feed the growing population while using fewer resources.

Myopathies

Over the past several decades, the poultry industry has had an emphasis on production efficiency, health, and quantity of broiler meat, but the industry needs to focus on quality as well. There has been an unprecedented emergence of myopathies in recent years in most commercial broilers that have negative impacts on quality. These myopathies have been associated with rapid growth and high breast yield. Three major issues that have become most apparent in recent years are white striping (WS), woody (i.e., hardness) breast (WB), and spaghetti meat (SM). These myopathies are issues around the world that have major economic implications, based on 10 to 40% incidence of moderate/severe cases. Oftentimes, fillets can have more than one of the myopathies/defects present. Severe cases have resulted in unnecessary condemnations, decreased meat quality and yield, changed nutritional content, and continued reduced customer/consumer acceptance (resulting in lost customer accounts, leading to losses in the millions of dollars each year.

Various factors have been associated with woody breast in broilers, including metabolic and growth problems associated with large muscle fibers produced by fast-growing broilers, reductions in muscle stem cells, and satellite cell functions (activity and number). While there is ongoing research to determine the root causes of these myopathies, the focus has also been placed on mitigation strategies. These mitigation strategies should involve genetic and live production factors, including nutritional strategies. However, there is also a need for strategies in the processing plant. This includes quality-based sorting of meat to provide the highest quality end product to consumers, which requires identification and detection of quality issues.

Several methods of detection of the woody breast have been studied. Tactile evaluation and the use of instrumental compression force are the standards for detecting WB, but it can be labor intensive (Tijare et al., 2016; Caldas-Cueva and Owens, 2020). Manual tactile evaluation requires training of personnel and can be subjective at times, but is highly correlated to instrumental compression force (Sun et al., 2018). Near-infrared and hyperspectral imaging have had success by assessing proximate composition (Wold et al., 2017, 2019) where WB is associated with lower protein content. Yoon et al. (2020) developed an imaging system to detect fillet rigidity. This USDA ARS lab is continuing work to include other detection systems such as hyperspectral imaging (Yoon et al., 2016). Hanning et al. (2020) identified relationships between carcass shape and WB in deboned fillets with the goal to incorporate into existing vision grading systems.

Other novel technologies have also been studied, including accelerometers and bioelectrical impedance. However, while there are many ongoing efforts to develop innovative detection methods, yet relevant methods can continue to be used. Though high predictability rates are observed in research settings, the challenge is that woody breast is not homogeneous in nature. A combination of methods may be advantageous for better detectability in processing plants. The use of artificial intelligence may also allow for improvement in predictability.

Chilling Improvements for Improved Chilling and Food Safety. Chilling is an important step in poultry processing. It is critical to reduce carcasses temperatures in a timely manner, and it is a primary area where antimicrobials are used. With the increase in bird sizes over the recent years, there has been a subsequent increase in dwell time in the chiller. This uses valuable resources such as water, energy, and even space. Researchers have developed new chilling methods that would decrease chill time, thereby saving on resources and also improving food safety. The use of sub-zero saline chilling (SSC) has been studied as a means to chill carcasses faster and also for bacterial reduction. Kang and Hurley (2019) reported that dwell times in chillers were reduced as NaCl concentration increased and water temperature decreased. Based on previous studies, Kang et al. (2022) used SSC (4% NaCl/-2.41°C) and then combined it with a hot water spray (HWS) prior to chilling (71°C for 1 min). Using SSC alone or in combination with HWS resulted in significant bacterial reductions compared to those carcasses chilled in water immersion only or water immersion combined with HWS. This technology has good potential to reduce dwell times and improve food safety, potentially eliminating the use of antimicrobials in chillers. More research is needed to determine efficiency or process at full scale and quality effects.

Researchers from the Agricultural Technology Research Program (ATRP) at Georgia Agricultural Research Institute (GTRI) have developed other technologies for poultry chilling, including rotational movement in the chiller while on shackles and the use of fine ice slurry for the chilling medium (Britton, 2021). Initial results showed that rotational movement in the chiller combined with common line speeds could reduce dwell times by as much as 40% though more research is needed, especially for larger birds.

Innovation in Processing Plants

Labor is a critical need for the poultry industry, but in recent times, labor shortages have also been a major issue as a result of high turnover due to working environments (cold), repetitive and tedious tasks, and the pandemic. The development of innovative technologies in completing certain tasks can not only improve the use of labor (reducing the tedious tasks) but also provide a solution to labor shortages in some cases. Scientists have developed a virtual reality (VR) based operation to assist in broiler deboning operations at ATRP/GTRI (Britton, 2021). One such operation is cone loading, where a human would operate a robotic arm through the use of VR to load front halves onto a deboning line. The use of VR could be applied to tasks that have been previously hard to automate due to variability of the product (chicken) size and shape. There is potential for these operations to be directed by humans in a remote site rather than in the plant environment.

Deboning carcasses is a task that is performed manually or through automation. Automated deboners have been used in processing plants for well over 20 years, and their performance has improved over the years. However, manual debone lines typically still have better performance in terms of yield when compared to automated systems. Researchers at ATRP/GTRI have been working to develop better automated deboning systems. They have developed an intelligent cutting system that uses 3D imaging systems that communicate tasks to a robotic arm for deboning or cutting broiler carcasses. This imaging system, combined with collecting data on how humans cut (knife trajections) when deboning has allowed researchers to optimize knife path functions via machine learning. The combination of 3D imaging and robotics can allow for intelligent adjustment of the cutting process to account for natural variation from carcass to carcass. This allows for much-improved performance that can match human deboners.

Using hyperspectral imaging in poultry processing is another innovation for the poultry industry. Commercial systems are available for detecting foreign material and even product quality defects such as woody breast.

Food Safety

Food safety continues to be a priority area for poultry processors as it is critical to provide a safe food supply. Microorganisms such as Salmonella spp. and Camplyobacter spp. remain a focal point for food safety research. Evaluating these microorganisms from a system approach (live production through processing) is important as incoming microbiological loads to the processing plant can affect the loads in final products. In processing plants, it is necessary to develop new antimicrobials to be used on poultry products, as microbes can adapt to their environment. When choosing antimicrobials, it is also key to understand the impact on product quality. Often times, antimicrobials can have negative effects on quality, so, balancing quality and effectiveness against bacteria is important. There is much ongoing research to understand the microbiome and the impact of antimicrobial treatments in live animals or postharvest. Additionally, it is a priority to use the quantification of Salmonella in the processing plants as a method for assessing its risk for human health, rather than its presence or absence.