Integrating heterogeneous across-country data for proxy-based random forest prediction of enteric methane in dairy cattle.

Direct measurements of methane (CH4) from individual animals are difficult and expensive. Predictions based on proxies for CH4 are a viable alternative. Most prediction models are based on multiple linear regressions (MLR) and predictor variables that are not routinely available in commercial farms, such as dry matter intake (DMI) and diet composition. The use of machine learning (ML) algorithms to predict CH4 emissions from across-country heterogeneous data sets has not been reported. The objectives were to compare performances of ML ensemble algorithm random forest (RF) and MLR models in predicting CH4 emissions from proxies in dairy cows, and assess effects of imputing missing data points on prediction accuracy. Data on CH4 emissions and proxies for CH4 from 20 herds were provided by 10 countries. The integrated data set contained 43,519 records from 3,483 cows, with 18.7% missing data points imputed using k-nearest neighbor imputation. Three data sets were created, 3k (no missing records), 21k (missing DMI imputed from milk, fat, protein, body weight), and 41k (missing DMI, milk fat, and protein records imputed). These data sets were used to test scenarios (with or without DMI, imputed vs. nonimputed DMI, milk fat, and protein), and prediction models (RF vs. MLR). Model predictive ability was evaluated within and between herds through 10-fold cross-validation. Prediction accuracy was measured as correlation between observed and predicted CH4, root mean squared error (RMSE) and mean normalized discounted cumulative gain (NDCG). Inclusion of DMI in the model improved within and between-herd prediction accuracy to 0.77 (RMSE = 23.3%) and 0.58 (RMSE = 31.9%) in RF and to 0.50 (RMSE = 0.327) and 0.13 (RMSE = 42.71) in MLR, respectively than when DMI was not included in the predictive model. When missing DMI records were imputed, within and between-herd accuracy increased to 0.84 (RMSE = 18.5%) and 0.63 (RMSE = 29.9%), respectively. In all scenarios, RF models out-performed MLR models. Results suggest routinely measured variables from dairy farms can be used in developing globally robust prediction models for CH4 if coupled with state-of-the-art techniques for imputation and advanced ML algorithms for predictive modeling.

Comparison of Methods to Measure Methane for Use in Genetic Evaluation of Dairy Cattle.

Partners in Expert Working Group WG2 of the COST Action METHAGENE have used several methods for measuring methane output by individual dairy cattle under various environmental conditions. Methods included respiration chambers, the sulphur hexafluoride (SF6) tracer technique, breath sampling during milking or feeding, the GreenFeed system, and the laser methane detector. The aim of the current study was to review and compare the suitability of methods for large-scale measurements of methane output by individual animals, which may be combined with other databases for genetic evaluations. Accuracy, precision and correlation between methods were assessed. Accuracy and precision are important, but data from different sources can be weighted or adjusted when combined if they are suitably correlated with the 'true' value. All methods showed high correlations with respiration chambers. Comparisons among alternative methods generally had lower correlations than comparisons with respiration chambers, despite higher numbers of animals and in most cases simultaneous repeated measures per cow per method. Lower correlations could be due to increased variability and imprecision of alternative methods, or maybe different aspects of methane emission are captured using different methods. Results confirm that there is sufficient correlation between methods for measurements from all methods to be combined for international genetic studies and provide a much-needed framework for comparing genetic correlations between methods should these become available.

Prediction of enteric methane production, yield, and intensity in dairy cattle using an intercontinental database.

Enteric methane (CH4 ) production from cattle contributes to global greenhouse gas emissions. Measurement of enteric CH4 is complex, expensive, and impractical at large scales; therefore, models are commonly used to predict CH4 production. However, building robust prediction models requires extensive data from animals under different management systems worldwide. The objectives of this study were to (1) collate a global database of enteric CH4 production from individual lactating dairy cattle; (2) determine the availability of key variables for predicting enteric CH4 production (g/day per cow), yield [g/kg dry matter intake (DMI)], and intensity (g/kg energy corrected milk) and their respective relationships; (3) develop intercontinental and regional models and cross-validate their performance; and (4) assess the trade-off between availability of on-farm inputs and CH4 prediction accuracy. The intercontinental database covered Europe (EU), the United States (US), and Australia (AU). A sequential approach was taken by incrementally adding key variables to develop models with increasing complexity. Methane emissions were predicted by fitting linear mixed models. Within model categories, an intercontinental model with the most available independent variables performed best with root mean square prediction error (RMSPE) as a percentage of mean observed value of 16.6%, 14.7%, and 19.8% for intercontinental, EU, and United States regions, respectively. Less complex models requiring only DMI had predictive ability comparable to complex models. Enteric CH4 production, yield, and intensity prediction models developed on an intercontinental basis had similar performance across regions, however, intercepts and slopes were different with implications for prediction. Revised CH4 emission conversion factors for specific regions are required to improve CH4 production estimates in national inventories. In conclusion, information on DMI is required for good prediction, and other factors such as dietary neutral detergent fiber (NDF) concentration, improve the prediction. For enteric CH4 yield and intensity prediction, information on milk yield and composition is required for better estimation.

Exploring the methanogen and bacterial communities of rumen environments: solid adherent, fluid and epimural.

The rumen microbiome occupies a central role in animal health and productivity. A better understanding of the rumen ecosystem is essential to increase productivity or decrease methane production. Samples were collected from the three main rumen environments: the solid-adherent fraction, the liquid fraction and the epithelium. For the liquid and solid fraction, two alternative sample processing protocols were compared, resulting in a total of five sample types: crude solids (S), the eluted solid-adherent fraction (Ad), free-living species in the crude rumen liquid (CRL), strained liquid samples (Lq) and epimural scrapings (Ep). The bacterial and methanogen communities of these sample types were analysed using 16S metabarcoding and qPCR. The results indicate that the liquid and solid-adherent environments are distinguished mainly by the differential abundance of specific taxonomic groups. Cellulolytic bacteria that pioneer biofilm formation, together with secondary colonisers are prevalent in solid-adherent samples, while dominant species in the fluid samples are primarily identified as consumers of soluble nutrients. Also, methanogen species are found to have a preference for either a solid-adherent or free-living occurrence. The epimural environment is characterised by a different microbial profile. Ten bacterial families and two methanogen genera are almost exclusively found in this environment.

Proteomic analysis of the honey bee worker venom gland focusing on the mechanisms of protection against tissue damage.

Honey bee workers use venom for the defence of the colony and themselves when they are exposed to dangers and predators. It is produced by a long thin, convoluted, and bifurcated gland, and consists of several toxic proteins and peptides. The present study was undertaken in order to identify the mechanisms that protect the venom gland secretory cells against these harmful components. Samples of whole venom glands, including the interconnected reservoirs, were separated by two-dimensional gel electrophoresis and the most abundant protein spots were subjected to mass spectrometric identification using MALDI TOF/TOF-MS and LC MS/MS. This proteomic study revealed four antioxidant enzymes: CuZn superoxide dismutase (SOD1), glutathione-S-transferase sigma 1 isoform A (GSTS1), peroxiredoxin 2540 (PXR2540) and thioredoxin peroxidase 1 isoform A (TPX1). Although glutathione-S-transferase (GST) has also been associated with xenobiotic detoxification, the protein we found belongs to the GST Sigma class which is known to protect against oxidative stress only. Moreover, we could demonstrate that the GST and SOD activity of the venom gland was low and moderate, respectively, when compared to other tissues from the adult honey bee. Several proteins involved in other forms of stress were likewise found but it remains uncertain what their function is in the venom gland. In addition to major royal jelly protein 9 (MRJP9), already found in a previous proteomic study, we identified MRJP8 as second member of the MRJP protein family to be associated with the venom gland. Transcripts of both MRJPs were amplified and sequenced. Two endocuticular structural proteins were abundantly present in the 2D-gel and most probably represent a structural component of the epicuticular lining that protects the secretory cells from the toxins they produce.

Molecular cloning and expression of icarapin, a novel IgE-binding bee venom protein.

The 1045bp full-length cDNA sequence of a new bee venom component was obtained by rapid amplification of cDNA ends. The 672bp coding sequence corresponds to a protein with a signal peptide and multiple carbohydrate binding sites, and it was named icarapin. It has the new consensus sequence N-[TS]-T-S-[TV]-x-K-[VI](2)-[DN]-G-H-x-V-x-I-N-[ED]-T-x-Y-x-[DHK]-x(2,6)- [STA]-[VLFI]-x-[KR]-V-R-[VLI]-[IV]-[DN]-V-x-P. At least two transcript variants were found. Recombinant icarapin was tested for recognition by IgE antibodies and gave a positive dot blot with sera from 4 out of 5 bee venom allergic patients, all beekeepers. Indirect immunofluorescent staining localized the protein in the cuticular lining of the venom duct.

Identification of Paenibacillus larvae to the subspecies level: an obstacle for AFB diagnosis.

This study was initially aimed at developing a PCR-test to differentiate between the pathogenic agent of American foulbrood (Paenibacillus larvae subsp. larvae) and powdery-scale disease (P. larvae subsp. pulvifaciens) of the honeybee. The test was based on the "insert of clone 9" (iC9), referring to a cloned 1.9 kB HaeIII fragment that occurs only in the P. larvae subsp. larvae reference strains and possibly correlates with American foulbrood virulence. It was shown that an iC9-based PCR-test discriminates between the BCCM/LMG reference strains of both subspecies. However, the screening of 179 Belgian field strains revealed five isolates that gave no iC9-based amplicon, thus rather resembling to P. larvae subsp. pulvifaciens. In addition, they all produced acid from mannitol, a characteristic previously assigned to the pulvifaciens subspecies. Because the reference strains gave conflicting data, this carbohydrate acidification was not conclusive. Therefore, the exact taxonomic position of the five retained strains was determined by a polyphasic approach using SDS-PAGE, AFLP, and ERIC-based PCR. Four iC9-negative field strains could be identified as P. larvae subsp. larvae; the taxonomic position of the fifth field strain remained ambiguous. The latter was provisionally classified as a subspecies pulvifaciens strain on the basis of SDS-PAGE. The present paper demonstrates the existence of field strains that do not fit well in the subdivision of the species P. larvae into two subspecies. Knowing that only one of both subspecies represents the pathogenic agent of AFB, this is a serious obstacle for the diagnosis of this honeybee disease.

The protein composition of honeybee venom reconsidered by a proteomic approach.

Pure honeybee venom samples were submitted to two-dimensional gel electrophoresis. A total of 49 excised spots were analyzed by mass spectrometry; 39 of them resulted in the identification of 6 different known bee venom proteins and of 3 proteins that have not been described in such samples before. The first new venom protein has a platelet-derived and vascular endothelial growth factor family domain, the second protein shows no homologies with any known protein and the third matches a hypothetical protein similar to major royal jelly protein 8.