Assessment of Soybean Lodging Using UAV Imagery and Machine Learning.

Plant lodging is one of the most essential phenotypes for soybean breeding programs. Soybean lodging is conventionally evaluated visually by breeders, which is time-consuming and subject to human errors. This study aimed to investigate the potential of unmanned aerial vehicle (UAV)-based imagery and machine learning in assessing the lodging conditions of soybean breeding lines. A UAV imaging system equipped with an RGB (red-green-blue) camera was used to collect the imagery data of 1266 four-row plots in a soybean breeding field at the reproductive stage. Soybean lodging scores were visually assessed by experienced breeders, and the scores were grouped into four classes, i.e., non-lodging, moderate lodging, high lodging, and severe lodging. UAV images were stitched to build orthomosaics, and soybean plots were segmented using a grid method. Twelve image features were extracted from the collected images to assess the lodging scores of each breeding line. Four models, i.e., extreme gradient boosting (XGBoost), random forest (RF), K-nearest neighbor (KNN) and artificial neural network (ANN), were evaluated to classify soybean lodging classes. Five data preprocessing methods were used to treat the imbalanced dataset to improve classification accuracy. Results indicate that the preprocessing method SMOTE-ENN consistently performs well for all four (XGBoost, RF, KNN, and ANN) classifiers, achieving the highest overall accuracy (OA), lowest misclassification, higher F1-score, and higher Kappa coefficient. This suggests that Synthetic Minority Oversampling-Edited Nearest Neighbor (SMOTE-ENN) may be a good preprocessing method for using unbalanced datasets and the classification task. Furthermore, an overall accuracy of 96% was obtained using the SMOTE-ENN dataset and ANN classifier. The study indicated that an imagery-based classification model could be implemented in a breeding program to differentiate soybean lodging phenotype and classify lodging scores effectively.

Use of regression models for development of a simple and effective biogas decision-support tool.

Anaerobic digestion (AD) is an alternative way to treat manure while producing biogas as a renewable fuel. To increase the efficiency of AD performance, accurate prediction of biogas yield in different working conditions is necessary. In this study, regression models were developed to estimate biogas production from co-digesting swine manure (SM) and waste kitchen oil (WKO) at mesophilic temperatures. A dataset was collected from the semi-continuous AD studies across nine treatments of SM and WKO, evaluated at 30, 35 and 40 °C. Application of polynomial regression models and variable interactions with the selected data resulted in an adjusted R2 value of 0.9656, much higher than the simple linear regression model (R2 = 0.7167). The significance of the model was observed with the mean absolute percentage error score of 4.16%. Biogas estimation using the final model resulted in a difference between predicted and actual values from 0.2 to 6.7%, except for one treatment which was 9.8% different than observed. A spreadsheet was created to estimate biogas production and other operational factors using substrate loading rates and temperature settings. This user-friendly program could be used as a decision-support tool to provide recommendations for some working conditions and estimation of the biogas yield under different scenarios.

Optimization and microbial diversity of anaerobic co-digestion of swine manure with waste kitchen oil at high organic loading rates.

Anaerobic co-digestion of swine manure (SM) and waste kitchen oil (WKO) was conducted to evaluate the effect of high organic loading rates (OLRs) on biogas production efficiency and microbial changes. Combinations of different loading rates of SM and WKO, with total OLRs from 2 to 8 g VS (volatile solid)/L/d, were evaluated in a laboratory-scale study. While feeding more than 4 g VSSM/L/d did not result in higher biogas production in both mono- and co-digestion scenarios, the addition of WKO increased the total OLR up to 6 g VS/L/d without significant reduction of system productivity. Biogas yields of M2O1 (2 g VSSM/L/d + 1 g VSWKO/L/d) and M4O2 were 910 ± 35 and 849 ± 85 mL/g VSfed which were 25.2 % and 16.9 % higher than the mono-digestion of M2, respectively. A significant increase of bacterial alpha-diversity (Shannon index) was observed in M2O1, at 233.0 ± 3.6 compared with 218.7 ± 5.1 of M2 (p < 0.05). Less bacterial alpha-diversity and accumulation of volatile fatty acids were observed in M4O1 and M4O2, suggesting their potential instability. When digesters were fed with M2, the introduction of 1.4 g VSWKO/L/d or more did not increase biogas yield and could cause system imbalance. The study suggests the limit of WKO could be increased when higher OLRs of SM were applied but should not be more than 4 g VSSM/L/d, and ratio between SM and WKO should be considered to avoid failure. Some of the system disturbances took up to three months to show.

Dairy and swine manure management - Challenges and perspectives for sustainable treatment technology.

Global dairy and swine production growth has increased significantly over the past decades, resulting in higher manure generation in certain areas and environmental concerns. Therefore, manure management is an essential focus for farmers and environmental regulators. Systematic selection of manure management practices can provide environmental benefits, but accounting for local constraints, economics and farming practices are significant challenges. All these factors drive the selection of appropriate manure management systems (MMSs). MMSs are highly varied for their design, partly due to individual farm settings, geography, and the end-use applications of manure. However, the benefits of technological advancements in MMSs provide higher manure treatment efficiency and co-production of value-added products such as recycled water, fiber, sand bedding, and nutrient-rich bio-solids, among others. To achieve higher environmental benefits, advanced manure treatment technologies have to be implemented, which comes with additional costs. So, there is a tradeoff between environmental benefits and cost. With the above prospects, this article reviews: 1) the different treatment technologies used in dairy and swine farms, 2) the life cycle assessment (LCA) method's importance in evaluating various treatment technologies for better environmental returns, and 3) decision support tools (DST) and their significance in MMSs prioritization. We found considerable heterogeneity in the available datasets, mainly on crucial parameters such as water consumption, types and amount of bedding materials, manure removal frequency, manure treatment technologies, and the extent of resource recovery. Thus, suitable environmental impact assessment inventory models are needed to evaluate a more comprehensive range of treatment technologies in MMSs, representing the spatial and farming system heterogeneities. There is also a need for user-friendly DST with adjustable inputs for the functional components of MMSs and evaluation criteria, which can rapidly evaluate the techno-economic feasibility of alternative systems.

Degradation of Veterinary Antibiotics in Swine Manure via Anaerobic Digestion.

Antibiotic-resistant microorganisms are drawing a lot of attention due to their severe and irreversible consequences on human health. The animal industry is considered responsible in part because of the enormous volume of antibiotics used annually. In the current research, veterinary antibiotic (VA) degradation, finding the threshold of removal and recognizing the joint effects of chlortetracycline (CTC) and Tylosin combination on the digestion process were studied. Laboratory scale anaerobic digesters were utilized to investigate potential mitigation of VA in swine manure. The digesters had a working volume of 1.38 L (in 1.89-L glass jar), with a hydraulic retention time (HRT) of 21 days and a loading rate of 1.0 g-VS L-1 d-1. Digesters were kept at 39 ± 2 °C in incubators and loaded every two days, produced biogas every 4 days and digester pH were measured weekly. The anaerobic digestion (AD) process was allowed 1.5 to 2 HRT to stabilize before adding the VAs. Tests were conducted to compare the effects of VAs onto manure nutrients, volatile solid removal, VA degradation, and biogas production. Concentrations of VA added to the manure samples were 263 to 298 mg/L of CTC, and 88 to 263 mg/L of Tylosin, respectively. Analysis of VA concentrations before and after the AD process was conducted to determine the VA degradation. Additional tests were also conducted to confirm the degradation of both VAs dissolved in water under room temperature and digester temperature. Some fluctuations of biogas production and operating variables were observed because of the VA addition. All CTC was found degraded even only after 6 days of storage in water solution; thus, there was no baseline to estimate the effects of AD. As for Tylosin, 100% degradation was observed due to the AD (removal was 100%, compared with 24-40% degradation observed in the 12-day water solution storage). Besides, complete Tylosin degradation was also observed in the digestate samples treated with a mixture of the two VAs. Lastly, amplicon sequencing was performed on each group by using the 50 most variable operational taxonomic units (OTUs)s and perfect discriminations were detected between groups. The effect of administration period and dosage of VAs on Phyla Firmicutes Proteobacteria, Synergistetes and Phylum Bacteroides was investigated. These biomarkers' abundance can be employed to predict the sample's treatment group.

Long-Term Mesophilic Anaerobic Co-Digestion of Swine Manure with Corn Stover and Microbial Community Analysis.

Long-term anaerobic co-digestion of swine manure (SM) and corn stover (CS) was conducted using semi-continuously loaded digesters under mesophilic conditions. A preliminary test was first conducted to test the effects of loading rates, and results indicated the 3 g-VS L-1 d-1 was the optimal loading rate. Based on the preliminary results, a verification replicated test was conducted with 3 g-VS L-1 d-1 loading rate and different SM/CS ratios (1:1, 2:1 and 1:2). Results showed that a SM/CS ratio of 2/1 was optimal, based on maximum observed methane-VSdes generation and carbon conversion efficiency (72.56 ± 3.40 mL g-1 and 40.59%, respectively). Amplicon sequencing analysis suggested that microbial diversity was increased with CS loading. Amino-acid-degrading bacteria were abundant in the treatment groups. Archaea Methanoculleus could enhance biogas and methane productions.

Emissions monitoring at a deep-pit swine finishing facility: research methods and system performance.

This paper describes part of a comprehensive National Air Emissions Monitoring Study (NAEMS) conducted at a swine finishing farm located in the state of Indiana, in the United States. The NAEMS was a 2-year study of emissions from animal feeding operations that produce pork, chicken meat, eggs, and milk. It provided emission data for the US. Environmental Protection Agency (EPA) to develop tools for estimating emissions from livestock farms. The study in Indiana focused on quantifying and characterizing emissions of gases, particulate matter (PM), and volatile organic compounds (VOCs) from a swine finishing quad (four 1000-head rooms under one roof). Long-term continuous and quasi-continuous measurements were conducted with 157 on-line measurement variables using an array of instruments and sensors for gas and PM concentrations, fan operation, room static pressures, indoor temperature and humidity, animal activity and feeding times, and weather conditions. Pig inventory and weight, feed type and quantity, and manure accumulation and composition were also documented. Systematic tests of the measurement system were conducted. Monitoring methodologies, instrumentation applications, equipment maintenance, quality controls, and system performances are presented and can be used as a reference in assessing research quality and improving future environmental studies on livestock facilities.

Methane and carbon dioxide emission from two pig finishing barns.

Agricultural activities are an important source of greenhouse gases. However, comprehensive, long-term, and high-quality measurement data of these gases are lacking. This article presents a field study of CH(4) and CO(2) emission from two 1100-head mechanically ventilated pig (Sus scrofa) finishing barns (B1 and B2) with shallow manure flushing systems and propane space heaters from August 2002 to July 2003 in northern Missouri. Barn 2 was treated with soybean oil sprinkling, misting essential oils, and misting essential oils with water to reduce air pollutant emissions. Only days with CDFB (complete-data-full-barn), defined as >80% of valid data during a day with >80% pigs in the barns, were used. The CH(4) average daily mean (ADM) emission rates were 36.2 +/- 2.0 g/d AU (ADM +/- 95% confidence interval; animal unit = 500 kg live mass) from B1 (CDFB days = 134) and 28.8 +/- 1.8 g/d AU from B2 (CDFB days = 131). The CO(2) ADM emission rates were 17.5 +/- 0.8 kg/d AU from B1 (CDFB days = 146) and 14.2 +/- 0.6 kg/d AU from B2 (CDFB days = 137). The treated barn reduced CH(4) emission by 20% (P < 0.01) and CO(2) emission by 19% (P < 0.01). The CH(4) and CO(2) released from the flushing lagoon effluent were equivalent to 9.8 and 4.1% of the CDFB CH(4) and CO(2) emissions, respectively. The emission data were compared with the literature, and the characteristics of CH(4) and CO(2) concentrations and emissions were discussed.

Quality-assured measurements of animal building emissions: particulate matter concentrations.

Federally funded, multistate field studies were initiated in 2002 to measure emissions of particulate matter (PM) < 10 microm (PM10) and total suspended particulate (TSP), ammonia, hydrogen sulfide, carbon dioxide, methane, nonmethane hydrocarbons, and odor from swine and poultry production buildings in the United States. This paper describes the use of a continuous PM analyzer based on the tapered element oscillating microbalance (TEOM). In these studies, the TEOM was used to measure PM emissions at identical locations in paired barns. Measuring PM concentrations in swine and poultry barns, compared with measuring PM in ambient air, required more frequent maintenance of the TEOM. External screens were used to prevent rapid plugging of the insect screen in the PM10 preseparator inlet. Minute means of mass concentrations exhibited a sinusoidal pattern that followed the variation of relative humidity, indicating that mass concentration measurements were affected by water vapor condensation onto and evaporation of moisture from the TEOM filter. Filter loading increased the humidity effect, most likely because of increased water vapor adsorption capacity of added PM. In a single layer barn study, collocated TEOMs, equipped with TSP and PM10 inlets, corresponded well when placed near the inlets of exhaust fans in a layer barn. Initial data showed that average daily mean concentrations of TSP, PM10, and PM2.5 concentrations at a layer barn were 1440 +/- 182 microg/m3 (n = 2), 553 +/- 79 microg/m3 (n = 4), and 33 +/- 75 microg/m3 (n = 1), respectively. The daily mean TSP concentration (n = 1) of a swine barn sprinkled with soybean oil was 67% lower than an untreated swine barn, which had a daily mean TSP concentration of 1143 +/- 619 microg/m3. The daily mean ambient TSP concentration (n = 1) near the swine barns was 25 +/- 8 microg/m3. Concentrations of PM inside the swine barns were correlated to pig activity.

Odor and gas release from anaerobic treatment lagoons for swine manure.

Odor and gas release from anaerobic lagoons for treating swine waste affect air quality in neighboring communities but rates of release are not well documented. A buoyant convective flux chamber (BCFC) was used to determine the effect of lagoon loading rate on measured odor and gas releases from two primary lagoons at a simulated wind speed of 1.0 m s(-1). Concentrations of ammonia (NH3), hydrogen sulfide (H2S), carbon dioxide (CO2), sulfur dioxide (SO2), and nitric oxide (NO) in 50-L air samples were measured. A panel of human subjects, whose sensitivity was verified with a certified reference odorant, evaluated odor concentration, intensity, and hedonic tone. Geometric mean odor concentrations of BCFC inlet and outlet samples and of downwind berm samples were 168 +/- 44 (mean +/- 95% confidence interval), 262 +/- 60, and 114 +/- 38 OU(E) m(-3) (OU(E), European odor unit, equivalent to 123 microg n-butanol), respectively. The overall geometric mean odor release was 2.3 +/- 1.5 OU(E) s(-1) m(-2) (1.5 +/- 0.9 OU s(-1) m(-2)). The live mass specific geometric mean odor release was 13.5 OU(E) s(-1) AU(-1) (animal unit = 500 kg live body mass). Overall mean NH3, H2S, CO2 and SO2 releases were 101 +/- 24, 5.7 +/- 2.0, 852 +/- 307, and 0.5 +/- 0.4 microg s(-1) m(-2), respectively. Nitric oxide was not detected. Odor concentrations were directly proportional to H2S and CO2 concentrations and odor intensity, and inversely proportional to hedonic tone and SO2 concentration (P < 0.05). Releases of NH3, H2S, and CO2 were directly proportional (P < 0.05) to volatile solids loading rate (VSLR).