Use of an Ethanol Bio-Reﬁnery Product as a Soy Bean Alternative in Diets for Fast-Growing Meat Production Species: A Circular Economy Approach

: The recent conceptual pivot from bioethanol production to ethanol bioreﬁning has led to development of protein derived by fractionating the non-ethanol streams post fermentation within the plant. The aim of this study was to identify the effect of replacing dietary soy with corn-fermented protein (CFP) on performance of fast-growing meat species and the impact on the carbon footprint associated with the feed for each species. The study contains trials on 3 species, broiler, turkey and salmon. In trial one, 324 broiler chicks were allocated randomly to 36 pens distributed into 3 dietary treatments; control (0% CFP), 5% CFP and 10% CFP; for 35 days. In trial 2, 150 turkey poults were allocated to 3 treatments: control (0 CFP), 4% CFP and 8% CFP for 35 days. In trial 3, 525 Atlantic Salmon (starting weight 304 g ± 10.7 g) were raised in 15 saltwater tanks for 84 days with 5 treatments, control (0% CFP), 5% CFP, 10% CFP, 15% CFP and 20% CFP. Growth response, nutrient utilisation and carbon footprint were assessed in each trial. Replacement of soy with CFP showed limited differences in growth response and nutrient utilization but replacing soy bean meal with CFP at rate of 5%, 8% and 10% in broiler, turkey and salmon diets, respectively resulted in a 14% decrease in carbon footprint of diet manufacturing. This investigation shows coupling bioethanol production with poultry and salmon production represents a highly effective circular economy contributing to multiple UN Sustainable Development Goals.


Introduction
Poultry and fish are both increasing in popularity as protein sources in human diets. Fish consumption has increased by 122% between 1990 and 2018 [1] with an over 500% rise in aquaculture production globally over the same timescale. Salmon production is estimated to reach 2.7 million tonnes (MT) by 2021, compared to 0.8 MT in 2000 [2]. Similarly, poultry meat production is also increasing year on year and is predicted to have reached 137 MT in 2020 [1]. This rapid increase in the scale of meat production inevitably creates a conflicting global role for poultry and fish production: in rapidly increasing their contribution to UN Sustainable Development Goal (SDG) 2 (Zero hunger) and SDG 3 (Good health and wellbeing), the negative implications for SDG 12 (Responsible consumption and production) and SDG 13 (Climate action) are concurrently increased through increased resource use and pollution, respectively. Development of circular economies is an effective way to reduce the livestock and poultry pollution, improve the utilization efficiency of resources [3], and to balance economic development and environmental protection [3]. It is well recognised that the potential for extensive environmental impact from poultry production is substantial if a linear model is followed [4] and the key to sustainable development in animal production is conversion of potential waste and pollutants into used in animal feed, legislatures such as The Renewable Fuel Standard in the USA and The Renewable Transport Fuel Obligation in the UK have driven massive increases in bioethanol production from cereal (first generation bioethanol). In 2011, the estimated global production was around 113 billion litres [15]. Production of bioethanol is a exogenous enzyme and fermentation-based technology using Saccharomyces cerevisiae yeast to produce ethanol and a residual mash co-product known as 'whole stillage' which is subsequently decanted into a fibrous wet portion and a liquid component with only around 60-85 g/kg dry matter [16], 'thin stillage' that contains the majority of the yeast protein and soluble components. In traditional ethanol fermentation systems, the thin stillage is evaporated into a syrup, remixed with the wet grain and dried to form Distiller's Dried Grains with Solubles (DDGS). The evaporative drying of the co-product is an energydemanding, expensive necessity in order to remove all the "waste" material not required for the production of ethanol from the distillery which, if not removed, would congest the primary process of ethanol production.
DDGS is successfully used in ruminant feed [17] but the high fibre content decreases feed intake and limits nutrient utilization in fast growing species such as pigs, salmon, turkey and broiler (meat-type) chicken [18]. Therefore, the biorefining process associated with bioethanol production has been recently adapted to develop high protein, low fibre biorefinery co-products more suitable for non-ruminant meat species than DDGS [19]. The aim of this study is to assess the nutritional viability of partially replacement of SBM with corn-fermented protein (CFP) in the feed of fast growing meat species and to determine the impact of SBM replacement with CFP on the carbon footprint associated with the feed for each species.

Study Design
Three animal growth experiments were conducted to evaluate the impact of replacing soy in salmon, turkey and broiler chickens. Institutional and national guidelines for the care and use of animals were followed and all experimental procedures involving animals were approved by the local ethical review committees. For each growth experiments, diets were formulated to meet the need of the age and strain of species used, with partial replacement of high (>44%) protein soy bean meal (SBM) by a high (>50%) protein biorefinery co-product (Corn-Fermented Protein from a Maximized Stillage Co-products™ process, Fluid Quip Technologies, Cedar Rapids, IA, USA) as summarized in Table 1.

Broiler Study
Male, Ross 308 broilers (n = 324) were obtained from a commercial hatchery at day of hatch. Chicks were weighed individually and allocated to 0.64 m 2 floor pens in groups of 9 birds per pen. Pens contained clean wood shavings as bedding and two nipple drinkers per pen. There were three dietary treatments with CFP inclusions of 0, 4 and 8% (see Table 1 and Appendix A for more information), and these were each randomly allocated to 12 pens by block. Feed was available from 50 cm troughs in each pen ad libitum. The room was maintained at 31 • C and reduced based on Ross guidelines and bird behaviour to reach approximately 21 • C by D21. The lighting was increased from 23 h on D1 to 18 h by D6, and the 6 h of dark were maintained throughout the rest of the study. Diets were commercially formulated using a wheat, soy bean meal base and were produced by a commercial manufacturing facility (Target Feeds Ltd., Shropshire, UK). Diets were fed in two phases, starter crumb (D0-21) and grower pellet (D21-42), and all diets contained 0.5% TiO 2 as an inert marker for digestibility measures. Nitrogen content of the diets was determined using a combustion analyser (Dumatherm N Pro, Gerhardt Analytical Systems, Germany) then multiplied by 6.25 to derive crude protein content. Dry matter content and ether extractable fat content of diets were analyzed according to [20] (methods 930.15 and 945.16, respectively) was analysed by bomb calorimetry with sucrose as a standard [21]. Diet analysis is available in Appendix A, Table A1.
Birds and feed were weighed weekly on a pen basis to calculate bodyweight gain, feed intake and feed conversion ratio (FCR). On D42, two birds per pen were euthanized by cervical dislocation and butchered by a trained operative to assess carcass yield. Breast, thigh and drumstick (excluding skin) were collected and pooled to give an incomplete estimate of carcass meat yield. Excreta was collected per pen on D42, with a minimum of 10 g collected per pen, and dried in a forced air oven at 105 • C until constant weight. Dried, ground excreta and diets were analysed for titanium dioxide content by the spectrophotometric method described by [22]. Excreta was also analysed for nitrogen content as previously described. Nitrogen retention was calculated using the following equation: Statistical analysis was carried out using R [23]. Data outside 2 standard deviations from the mean for each variable were considered as outliers. Data of the study were analysed according to the following model: where Y is the response variable, µ is the overall mean, Diet is the effect of the experimental diet and ε is the residual. The pen was used as an experimental unit. Means were separated using Tukey's test at p ≤ 0.05.

Turkey Study
Male, BUT6 turkey poults (n = 150) were collected on day of hatch and allocated in groups of five to 10 pens per treatment. Pens were 0.64 m 2 and contained two water nipples and clean wood shavings as bedding. The room was maintained at 32 • C from D0 and the temperature was reduced daily to reach 21 • C by D21. Lighting was 23 h of light a day until D2, decreasing by 1 h a day to reach 8 h of darkness, which was then maintained throughout the study. Diets were formulated by a commercial nutritionist to be matched for protein, amino acids and energy (formulations and analysis in Appendix A). Treatments contained 0, 4 and 8% of CFP as detailed in Table 1 (see Appendix A, Table A2 for diet details) and all diets contained 0.5% TiO 2 as an indigestible marker. Each treatment was randomly allocated by block to 10 pens and was feed and water were available ad libitum throughout the study. Diets were provided as starter, sieved crumb (D0-21) and grower, short pellet (D21-42) and manufactured by a commercial mill (Target Feeds, Ltd., Shropshire, UK). Diets were analysed as previously described.
Poults and feed were weighed weekly by pen to calculate bodyweight gain, feed intake and FCR. On D42 3 birds per pen were culled by cervical dislocation and ileal digesta collected from the region between Meckel's Diverticulum and the ileal cecal junction. Ileal digesta was pooled per pen and freeze dried and ground before analysis.
Diets and digesta were analysed for nitrogen content and titanium dioxide marker to assess ileal nitrogen digestibility using the following equation Statistical analysis was carried out using SPSS v.24. After KS testing to confirm normality, data were analysed using one way ANOVA to investigate the effect of dietary treatment on FCR, feed intake (per bird) and individual bodyweight gain for each weigh period of the study, and cumulatively. Where appropriate, Bonferroni post hoc testing was used to elucidate differences between diets/treatments.

Salmon Study
Atlantic Salmon (n = 525; starting weight 304 g ± 10.7 g) were raised in 15 saltwater (25 ppt) tanks with 35 fish per tank for 84 days. Tanks were maintained in a recirculating aquaculture system using 750 L tanks and temperatures maintained at 14.2 • C (±0.6 • C) with greater than 90% oxygen saturation. A control diet was formulated to meet the needs of Atlantic salmon [24] and four further diets were formulated to contain graded levels of CFP; 5, 10, 15 and 20%. Diets were manufactured and extruded via a twin-screw cooling extruder and vacuum coated in the oil component of the diet (see Table 1 and Appendix A, Table A3). Each diet was fed to three tanks of fish. Salmon were fed by hand to satiety three times a day.
Fish were weighed on D0, 28, 56 and 84 of the study, with 5 fish per tank used to provide a mean fish weight. Five fish per tank were euthanized on D84 for whole body protein composition via the AOAC method (990.03). These results were used to calculate the rate of protein deposition using the following equation [25]: where D j is deposition rate [mg ( • C • d)-1] of nutrient j, F j and I j are final and initial whole-body mass of nutrient j (mg) at the end and the beginning of the 84-day period, respectively, n stands for the day number covering the period from F j to I j , T i ( • C) is mean daily water temperature for day T i , the product of which results in units of degree-days. The results were analysed using one-way ANOVA and Tukey's multiple comparison test with GraphPad Prism version 8.1.2 for Windows. Statistical differences were considered significant at p < 0.05.

Carbon Footprint Calculation
Global Food LCA Institute (GFLI) database (version 28-Dec-2018) was used to calculate the carbon footprint of the experimental feed. The carbon footprint of feeds was calculated using ReCiPe 2016 midpoint (H) assessment method with "economic" method of allocation. The GLFI database does not contain the footprint of high protein, ethanol bio-refinery product. Thus, the carbon footprint of high protein, ethanol bio-refinery product was obtained from Tallentire et al. (2018). The carbon footprint (kg CO 2 e/kg feed) was normalized for 1 kg of live weigh growth of Turkey and Salmon and 1 kg of meat yield of broilers. Table 2 summarizes the performance of the broiler study and the nitrogen retention measured in the birds at D42. Feed intake increased with increasing levels of CFP in the diets, with intake of 10% CFP-fed birds being significantly higher than the control. FCR was also significantly increased for the 10% CFP diet. Nitrogen retention was significantly improved for the 5% CFP diet over both the control and the 10% diet.  Table 3 shows the calculated carbon emissions per kg of bird growth, and also per kg of meat production (based on carcass yield of breast, thigh and drum only). It can be seen that all three GHG calculations show a reduction in CO 2 emissions with increasing dietary content of CFP.

Turkey Study
The performance of the turkey poults at D42 is shown in Table 4. Bodyweights at D42 increased numerically with CFP inclusion, but there were no significant differences in any performance measure over the six-week trial period. Nitrogen retention was significantly increased at 10% dietary inclusion of CFP over the control, soy-based diet.

Salmon Study
The fish performance measures and the protein retention for the trial period (84 days) is shown in Table 6. Final salmon weights were significantly higher for the 10% CFP inclusion diet compared with the 20% inclusion diet (752.1 g vs. 663.7 g, respectively). Feed intake was also highest for the 10% CFP diet and the control diet compared with the 20% CFP diet. There were no significant differences in FCR or protein retention during the duration of this study.  Table 7 shows the calculated GHG emissions and CO 2 output for the salmon study. CO 2 output is reported on a per kg feed basis and also per kg of fish growth. Increasing inclusion of CFP reduces GHG emissions in a linear fashion.

Discussion
Previous studies theoretically modelling LCA have suggested that a CFP type product increases nitrogen excretion associated negative environmental impacts [26], but this study showed that nitrogen utilization was significantly improved with CFP for both poultry studies studied and not affected in salmon. Therefore, although the CFP has higher total protein nitrogen than SBM, its improved digestibility mitigates the additional feed nitrogen and reduces excretion to a similar level to SBM. Beyond the positive environmental impact associated with this improved nutrient utilization, further quantifiable effects are incurred earlier in the meat production process; associated with production of the individual feed material. Agricultural land use (ALU) associated with feed material production substantially contributes to CF, so maintaining low ALU values is a key focus in low carbon meat production. In the case of CFP, the ALU comes almost entirely from the production of the cereal crop [26]. However, with multiple product streams deriving from the fermentation process, this ALU tariff is spread across a number of co-products including as corn oil and bioethanol. Therefore, the calculated CF of diets including CFP are substantially reduced compared to diets relying on SBM as the main dietary protein source.
Use of soy as livestock feed outside the Americas incur a high CO 2 cost relating to the long distance of transportation. In addition, South American soy production is associated with high level of deforestation, resulting in particularly high ALU tariffs derived from the additional land use change. Accordingly, decreasing the dependency on soy in livestock nutrition would reduce the negative consequences of soy production and transportation on the environment due to the large penalties associated with land use change. CFP is currently produced at 500 thousand tonnes per annum from six bioethanol plants in the USA, which will increase to an estimated 1 million tonnes a year by the end of 2023. CFP is a co-product with high protein content which is produced from bioethanol generation. Accordingly, it is expected to be a low-GHG replacement of soy in livestock nutrition. Furthermore, it is more cost-effective source of nutrients for livestock compared to conventional feedstuffs [27].
The results of the current study show N retention of broilers fed diets containing 5% CFP in place of SBM, and 8% CFP in turkey diets was higher than the control; which conflicts with previous studies on traditional bioethanol co-product: DDGS. Ref. [28,29] reported that increased manure production and manure N excretion was produced by broilers fed high-protein corn distillers dried grains. This negative effect may be due to the increase in dietary fibre and reduction in protein digestibility resulting from heat damage associated with DDGS production [30]. However, the replacement of soy by CFP in the current study did not introduce sufficient fibre or heat-damaged protein to negatively impact on broiler performance or nitrogen retention. Furthermore, the high digestibility of CFP reduces N excretion in comparison to wholly SBM-based diets. This improvement in N retention of broilers fed 5% CFP would decrease the NH3 emission from broiler production, creating added value from 5% CFP inclusion beyond the quantified parameters reported in the current study. Interestingly, 10% dietary inclusion of CFP improved nitrogen retention of broilers, but also increased feed intake without a concurrent improvement in weight gain, so deleteriously effecting feed conversion ratio and therefore negatively impacting the economic viability of including 10% CFP in broiler diets. However, a 10% dietary inclusion of CFP provides a 19% reduction in CO 2 output compared with the control diet on a basis of per kg growth, and a 22% reduction in CO 2 emissions on the basis of kg of meat produced. This reduction is 11% (per kg meat) and 14% (per kg meat) for the 5% inclusion of CFP. Turkeys may be fed diets including to 8% CFP in place of SBM with no effect on performance but a reduction in GHG emissions of 14% compared to turkeys fed the control diet. Distillers dried grains with solubles were included in broilers (up to 12%) and Turkeys (up to 8% diets) without negative consequences on growth performance [30]. Similarly to the poultry studies, the salmon study showed no improvement in growth related to dietary SBM replacement with FP, and, as with the broilers, the highest CFP level diet led to a small, negative effect on growth. This lack of improved growth response in salmon is surprising as previous studies show DDGS may totally replaces fish meal in fish diets when fishmeal is included at a level of 12% of the diet [31]. The DDGS was used as a protein source in rainbow trout diet without negative effect on digestibility and growth [30]. It has also been reported that DDGS can be used at levels up to 90% of winter diets for channel catfish without amino acids supplementation [32,33]. In alignment with the poultry trials, inclusion of 10% CFP in salmon diets leads to a reduction in CO 2 cost of one kg growth of almost 14% over the control diet.
In all three species assessed, the impacts on growth performance were limited but the positive impacts of including CFP in place of soy on the CF of each species diet were substantial. This initial evidence that partial CFP replacement of dietary soy reduces the carbon footprint of meat production justifies further in vivo studies directly assessing carbon cost of meat from fish and poultry fed CFP-containing diets as predictive modelling approaches (Tallentyre et al., 2018) provide conflicting results.
In all three species assessed, the impacts on growth performance were limited but the positive impacts of including CFP in place of soy on the CF of each species diet were substantial. This initial evidence that partial CFP replacement of dietary soy reduces the carbon footprint of meat production justifies further in vivo studies directly assessing carbon cost of meat from fish and poultry fed CFP-containing diets as predictive modelling approaches (Tallentyre et al., 2018) provide conflicting results. Inclusion of CFP in broiler, turkey and salmon diets at a rate of 5%, 8% and 10%, respectively improved nitrogen retention while decreasing GHG emissions. This indicates partial replacement of soy with CFP in the diets of fast-growing meat species would reduce the environmental impact of meat production without impacting on growth performance.
Successful circular economies rely on precise alignment of needs between producers and users. Cereal-based bioethanol plants have historically produced a secondary product of low and inconsistent nutrient value that has limited attraction as an animal feed mate-rial. The expansion of cereal-based bioethanol production raised initial concerns that the concurrent increased supply of the traditional co-product (DDGS) would exceed feed use potential [34]. However, the reframing of bioethanol production into biorefining, where multiple product streams are empirically scrutinized and modelled for optimum plant design [35], has revolutionized the sustainability of cereal-based bioethanol plants. The new engineering and plant design focus has been on optimizing the generation of high-quality protein from the bioethanol plants [36]. The multiple-species evaluation of CFP reported here shows that the biorefinery approach has created a protein product aligned to the needs of very high-volume users: salmon and poultry meat producers. The economic impact on the bioethanol plants pivoting to a biorefinery approach with multiple high value streams was particularly apparent as demand for transport fuel decreased during the early phase COVID-19 pandemic [37]. The environmental impact of partially replacing soy with a biorefinery product (CFP) in the diet of salmon and poultry has been clearly demonstrated in the reported studies. This shows that the development of circular economies is not only an effective way to reduce the livestock and poultry pollution, but may also be used to improve the utilization efficiency of resources and support environmental protection, thereby allowing meat production to simultaneously supporting a number of UN SDGs without concurrent detriment to others.

Conclusions
Inclusion of CFP in broiler, turkey and salmon diets at a rate of 5%, 8% and 10%, respectively improved nitrogen retention while decreasing GHG emissions. This indicates partial replacement of soy with CFP in the diets of fast-growing meat species would reduce the environmental impact of meat production without impacting on growth performance. Corn-fermented protein (CFP) presents a nutritionally viable option for partial replacement of soy in fish and poultry feed. The replacement of 10% of dietary soy with CFP is associated with a 19% and 14% reduction in CO 2 production associated with poultry and salmon production, respectively. Coupling bioethanol production with poultry and salmon production represents a highly effective circular economy.

Data Availability Statement:
The datasets generated during the current study are available from the corresponding author on reasonable request.

Acknowledgments:
The authors would like to thank the research teams involved at Nottingham Trent University and The Center for Aquaculture Technologies Canada.

Conflicts of Interest:
Peter Williams is employed as a consultant by Fluid Quip Technologies, LLC and contributed to the design of the studies but not the interpretation of the findings, which was undertaken by the NTU authors who declare no conflict of interest.