Mid-infrared Prediction Research Articles

Whole-genome scan to detect quantitative trait loci associated with milk protein composition in 3 French dairy cattle breeds.

In the context of the PhénoFinLait project, a genome-wide analysis was performed to detect quantitative trait loci (QTL) that affect milk protein composition estimated using mid-infrared spectrometry in the Montbéliarde (MO), Normande (NO), and Holstein (HO) French dairy cattle breeds. The 6 main milk proteins (α-lactalbumin, β-lactoglobulin, and αS1-, αS2-, β-, and κ-caseins) expressed as grams per 100g of milk (% of milk) or as grams per 100g of protein (% of protein) were estimated in 848,068 test-day milk samples from 156,660 cows. Genotyping was performed for 2,773 MO, 2,673 NO, and 2,208 HO cows using the Illumina BovineSNP50 BeadChip (Illumina Inc., San Diego, CA). Individual test-day records were adjusted for environmental effects and then averaged per cow to define the phenotypes analyzed. Quantitative trait loci detection was performed within each breed using a linkage disequilibrium and linkage analysis approach. A total of 39 genomic regions distributed on 20 of the 29 Bos taurus autosomes (BTA) were significantly associated with milk protein composition at a genome-wide level of significance in at least 1 of the 3 breeds. The 9 most significant QTL were located on BTA2 (133 Mbp), BTA6 (38, 47, and 87 Mbp), BTA11 (103 Mbp), BTA14 (1.8 Mbp), BTA20 (32 and 58 Mbp), and BTA29 (8 Mbp). The BTA6 (87 Mbp), BTA11, and BTA20 (58 Mbp) QTL were found in all 3 breeds, and they had highly significant effects on κ-casein, β-lactoglobulin, and α-lactalbumin, expressed as a percentage of protein, respectively. Each of these QTL explained between 13% (BTA14) and 51% (BTA11) of the genetic variance of the trait. Many other QTL regions were also identified in at least one breed. They were located on 14 additional chromosomes (1, 3, 4, 5, 7, 15, 17, 19, 21, 22, 24, 25, 26, and 27), and they explained 2 to 8% of the genetic variance of 1 or more protein composition traits. Concordance analyses, performed between QTL status and sequence-derived polymorphisms from 13 bulls, revealed previously known causal polymorphisms in LGB (BTA11) and GHR (BTA20 at 32 Mbp) and excluded some other previously described mutations. These results constitute a first step in identifying causal mutations and using routinely collected mid-infrared predictions in future genomic selection programs to improve bovine milk protein composition.

On the role of mid-infrared predicted phenotypes in fertility and health dairy breeding programs

Fertility and health traits are of prime importance in dairy breeding programs. However, these traits are generally complex, difficult to record, and lowly heritable (<0.10), thereby hampering genetic improvement in disease resistance and fertility. Hence, indicators are useful in the prediction of genetic merit for fertility and health traits as long as they are easier to measure than direct fitness traits, heritable, and genetically correlated. Considering that changes in (fine) milk composition over a lactation reflect the physiological status of the cow, mid-infrared (MIR) analysis of milk opens the door to a wide range of potential indicator traits of fertility and health. Previous studies investigated the phenotypic and genetic relationships between fertility and MIR-predicted phenotypes, most being related to negative postpartum energy balance and body fat mobilization (e.g., fat:protein ratio, urea, fatty acids profile). Results showed that a combination of various fatty acid traits (e.g., C18:1 cis-9 and C10:0) could be used to improve fertility. Furthermore, occurrence of (sub)clinical ketosis has been related to milk-based phenotypes such as fat:protein ratio, fatty acids, and ketone bodies. Hence, MIR-predicted acetone and β-hydroxybutyrate contents in milk could be useful for breeding cows less susceptible to ketosis. Although studies investigating the genetic association among mastitis and MIR-predicted phenotypes are scarce, a wide range of traits, potentially predicted by MIR spectrometry, are worthy of consideration. These include traits related to the disease response of the cow (e.g., lactoferrin), reduced secretory activity (e.g., casein), and the alteration of the blood-milk barrier (e.g., minerals). Moreover, direct MIR prediction of fertility and health traits should be further considered. To conclude, MIR-predicted phenotypes have a role to play in the improvement of dairy cow fertility and health. However, further studies are warranted to (1) grasp underlying associations among MIR-predicted indicator and fitness traits, (2) estimate the genetic parameters, and (3) include these traits in broader breeding strategies.

Technical note: Improving the accuracy of mid-infrared prediction models by selecting the most informative wavelengths

Mid-infrared spectroscopy (MIRS) is widely used to collect milk phenotypes at the population level. The aim of this study was to test the ability of the uninformative variable elimination (UVE) method to select and remove uninformative wavelength variables before partial least squares (PLS) analysis. Milk titratable acidity (TA) and Ca content were used as examples to illustrate the procedure. Reference values and MIRS spectra (n=208) of TA and Ca were retrieved from an existing database. The data set was randomly divided into calibration (70% of data) and validation (30% of data) sets, and PLS analysis was carried out before and after the UVE procedure. The UVE procedure selected 244 and 113 informative wavelengths for TA and Ca, respectively, from a total of 1,060. The elimination of uninformative variables before PLS regression increased the accuracy of MIRS prediction models, and it substantially reduced the computation time. Dealing with fewer variables is expected to enhance the efficiency of MIRS models to predict phenotypes at population level.

Feasibility study on the use of attenuated total reflectance mid-infrared spectroscopy for the analysis of malt quality parameters in wort

The feasibility of using attenuated total reflectance in combination with mid-infrared (MIR) spectroscopy for the analysis of malt quality parameters in wort, such as free α-amino acid (FAN), apparent attenuation limit (AAL), soluble protein (SP) and viscosity (VIS), was evaluated. The coefficients of determination in cross-validation RCV2 and the standard error of cross-validation were 0.81 and 10.3mgL-1 for FAN, 0.80 and 1.49% for AAL, 0.78 and 0.18% for SP and 0.69 and 0.13cP for VIS. Good calibration and validation statistics were obtained for AAL. However, for nitrogenous compounds the MIR predictions can only be used as approximate estimates of the true concentrations, for example to rank samples (low, medium, high) in the context of a breeding programme.

How useful are MIR predictions of total, particulate, humus, and resistant organic carbon for examining changes in soil carbon stocks in response to different crop management? A case study

Measures of particulate organic carbon (POC), humus organic carbon (HOC), and resistant organic carbon (ROC) (primarily char) are often used to represent the active, slow, and inert carbon pools used in soil carbon models. However, these fractions are difficult to measure directly, and mid infrared (MIR) spectroscopic techniques are increasingly being investigated to quantify these fractions and total organic carbon (TOC). This study examined the change in MIR-predicted pools of TOC, POC, HOC, and ROC in response to different crop management between two time periods (1981 and 2008) in a long-term wheat cropping trial in Queensland, Australia. The aims were (i) to assess the ability of MIR to detect changes in carbon stocks compared with direct measurements of TOC (LECO-TOC); and (ii) to assess how well the behaviour of POC, HOC, and ROC corresponded with the active, slow, and inert conceptual carbon pools. Significant declines in carbon stocks were observed over time using both LECO-TOC and MIR-predicted stocks of TOC, POC, HOC, and ROC, although MIR-TOC under-estimated loss by 27–30% compared with LECO-TOC. The decline in MIR-POC and MIR-HOC was consistent with the expected behaviour of the active and slow conceptual pools; however, the decline in ROC was not consistent with that of the inert pool. In addition, MIR measurements did not accurately detect differences in the rate of carbon loss under different crop management practices.

Combining management based indices with environmental parameters to explain regional variation in soil carbon under dryland cropping in South Australia

Identifying drivers of variation in soil organic carbon (OC) at a regional scale is often hampered by a lack of historical management information. Focusing on red-brown-earth soils (Chromosol) under dryland agriculture in the Mid-North and Eyre Peninsula of South Australia, our aims were 2-fold: (i) to provide a baseline of soil OC stocks (0.3 m) and OC fractions (mid-infrared predictions of particulate, humus, and resistant OC in 0.1 m samples) in cropping and crop-pasture systems; and (ii) to evaluate whether the inclusion of management-based indices could assist in explaining regional-level variation in OC stocks and fractions. Soil OC stocks in both regions varied ~20 Mg ha–1, with higher OC stocks in the Mid-North (38 Mg ha–1) than the Eyre Peninsula (29.1 Mg ha–1). The humus OC fraction was the dominant fraction, while the particulate OC was the most variable. Environmental variables only partially explained soil OC variability, with vapour pressure deficit (VPD) offering the greatest potential and likely acting as an integrator of temperature and moisture on plant growth and decomposition processes. Differences between broad-scale cropping and crop–pasture systems were limited. In the Mid-North, variability in soil OC stocks and fractions was high, and could not be explained by environmental or management variables. Higher soil OC concentrations (0.1 m) in the Eyre Peninsula cropping than crop–pasture soils were largely accounted for in the particulate OC fraction and are therefore unlikely to represent a long-term stable OC pool. Use of the management data in index format added some explanatory power to the variability in OC stocks over the main environmental variables (VPD, slope) within the Eyre Peninsula cropping soils only. In the wider context, the management data were useful in interpreting differences between regional findings and highlighted difficulties in using uninformed, broad-scale management categories.

Genetic variability of the mid-infrared prediction of lactoferrin content in milk for Walloon Holstein first-parity cows

The objective of this study was to assess the genetic variability of the mid-infrared prediction of lactoferrin content in milk (pLF) in Holstein first-parity cows. Variance components were estimated by Average Information Restricted Maximum Likelihood using a single-trait test-day random regression animal model. The dataset included 395,287 test-day records from 67,178 cows in 1190 herds from the Walloon Region of Belgium. Average pLF was 164.89mg/L and the standard deviation was 76.07mg/L. Frequency distribution for pLF was slightly asymmetrical, and pLF seemed to increase almost linearly all along the first lactation after a sharp decrease in early lactation. Genetic variance of pLF increased with days in milk within lactation while the permanent environmental variance was the highest in early lactation, then decreased to become lower than genetic variance at 50 days in milk, and finally increased in the last lactation stages. The pLF was a moderately heritable trait. Daily heritability of pLF was the lowest at 5 days in milk (0.19), then increased to reach a maximum at 260 days in milk (0.44), and finally decreased for the last stages of lactation (0.35 at 365 days in milk). Results from this study indicated that pLF is variable and heritable over the lactation and therefore it could be changed by genetic selection.

Mid-infrared prediction of lactoferrin content in bovine milk: potential indicator of mastitis

Lactoferrin (LTF) is a milk glycoprotein favorably associated with the immune system of dairy cows. Somatic cell count is often used as an indicator of mastitis in dairy cows, but knowledge on the milk LTF content could aid in mastitis detection. An inexpensive, rapid and robust method to predict milk LTF is required. The aim of this study was to develop an equation to quantify the LTF content in bovine milk using mid-infrared (MIR) spectrometry. LTF was quantified by enzyme-linked immunosorbent assay (ELISA), and all milk samples were analyzed by MIR. After discarding samples with a coefficient of variation between 2 ELISA measurements of more than 5% and the spectral outliers, the calibration set consisted of 2499 samples from Belgium (n = 110), Ireland (n = 1658) and Scotland (n = 731). Six statistical methods were evaluated to develop the LTF equation. The best method yielded a cross-validation coefficient of determination for LTF of 0.71 and a cross-validation standard error of 50.55 mg/l of milk. An external validation was undertaken using an additional dataset containing 274 Walloon samples. The validation coefficient of determination was 0.60. To assess the usefulness of the MIR predicted LTF, four logistic regressions using somatic cell score (SCS) and MIR LTF were developed to predict the presence of mastitis. The dataset used to build the logistic regressions consisted of 275 mastitis records and 13 507 MIR data collected in 18 Walloon herds. The LTF and the interaction SCS × LTF effects were significant (P < 0.001 and P = 0.02, respectively). When only the predicted LTF was included in the model, the prediction of the presence of mastitis was not accurate despite a moderate correlation between SCS and LTF (r = 0.54). The specificity and the sensitivity of models were assessed using Walloon data (i.e. internal validation) and data collected from a research herd at the University of Wisconsin – Madison (i.e. 5886 Wisconsin MIR records related to 93 mastistis events – external validation). Model specificity was better when LTF was included in the regression along with SCS when compared with SCS alone. Correct classification of non-mastitis records was 95.44% and 92.05% from Wisconsin and Walloon data, respectively. The same conclusion was formulated from the Hosmer and Lemeshow test. In conclusion, this study confirms the possibility to quantify an LTF indicator from milk MIR spectra. It suggests the usefulness of this indicator associated to SCS to detect the presence of mastitis. Moreover, the knowledge of milk LTF could also improve the milk nutritional quality.

Mid-infrared prediction of bovine milk fatty acids across multiple breeds, production systems, and countries

Increasing consumer concern exists over the relationship between food composition and human health. Because of the known effects of fatty acids on human health, the development of a quick, inexpensive, and accurate method to directly quantify the fatty acid (FA) composition in milk would be valuable for milk processors to develop a payment system for milk pertinent to their customer requirements and for farmers to adapt their feeding systems and breeding strategies accordingly. The aim of this study was (1) to confirm the ability of mid-infrared spectrometry (MIR) to quantify individual FA content in milk by using an innovative procedure of sampling (i.e., samples were collected from cows belonging to different breeds, different countries, and in different production systems); (2) to compare 6 mathematical methods to develop robust calibration equations for predicting the contents of individual FA in milk; and (3) to test interest in using the FA equations developed in milk as basis to predict FA content in fat without corrections for the slope and the bias of the developed equations. In total, 517 samples selected based on their spectral variability in 3 countries (Belgium, Ireland, and United Kingdom) from various breeds, cows, and production systems were analyzed by gas chromatography (GC). The samples presenting the largest spectral variability were used to calibrate the prediction of FA by MIR. The remaining samples were used to externally validate the 28 FA equations developed. The 6 methods were (1) partial least squares regression (PLS); (2) PLS+repeatability file (REP); (3) first derivative of spectral data+PLS; (4) first derivative+REP+PLS; (5) second derivative of spectral data+PLS; and (6) second derivative+REP+PLS. Methods were compared on the basis of the cross-validation coefficient of determination (R2cv), the ratio of standard deviation of GC values to the standard error of cross-validation (RPD), and the validation coefficient of determination (R2v). The third and fourth methods had, on average, the highest R2cv, RPD, and R2v. The final equations were built using all GC and the best accuracy was observed for the infrared predictions of C4:0, C6:0, C8:0, C10:0, C12:0, C14:0, C16:0, C18:0, C18:1 trans, C18:1 cis-9, C18:1 cis, and for some groups of FA studied in milk (saturated, monounsaturated, unsaturated, short-chain, medium-chain, and long-chain FA). These equations showed R2cv greater than 0.95. With R2cv equal to 0.85, the MIR prediction of polyunsaturated FA could be used to screen the cow population. As previously published, infrared predictions of FA in fat are less accurate than those developed from FA content in milk (g/dL of milk) and no better results were obtained by using milk FA predictions if no corrections for bias and slope based on reference milk samples with known contents of FA were used. These results indicate the usefulness of equations with R2cv greater than 95% in milk payment systems and the usefulness of equations with R2cv greater than 75% for animal breeding purposes.

Impact of fatty acid composition on the accuracy of mid-infrared fat analysis of farm milks

Our objective was to determine whether data from a previous study using model milk emulsions to characterize the influence of variation in fatty acid chain length and unsaturation on mid-infrared (MIR) fat predictions could be used to identify a strategy to improve the accuracy of MIR fat predictions on a population of farm milks with a wide variation in fatty acid chain length and unsaturation. The mean fatty acid chain length for 45 farm milks was 14.417 carbons, and the mean unsaturation was 0.337 double bonds per fatty acid. The range of fatty acid chain lengths across the 45 farm milks was 1.23 carbons, and the range in unsaturation was 0.167 double bonds per fatty acid. Fat B (absorbance by the carbon-hydrogen stretch) MIR predictions increased and fat A MIR (absorbance by the ester carbonyl stretch) predictions decreased relative to reference chemistry with increasing fatty acid chain length. When the fat B MIR fat predictions were corrected for sample-to-sample variation in unsaturation, the positive correlation between fat B and fatty acid chain length increased from a coefficient of determination of 0.42 to 0.89. A 45:55 ratio of fat B corrected for unsaturation and fat A gave a smaller standard deviation of the difference between MIR prediction and reference chemistry than any ratio of the fat B (without correction for unsaturation) and fat A or either fat B or fat A alone. This demonstrates the technical feasibility of this approach to improve MIR testing accuracy for fat, if a simple procedure could be developed to determine the unsaturation of fat in milk rapidly and to correct the fat B reading for the effect of unsaturation before being combined with fat A.

Influence of fatty acid chain length and unsaturation on mid-infrared milk analysis

Our first objective was to optimize center wavelengths and bandwidths for virtual filters used in a Fourier transform mid-infrared (MIR) milk analyzer. Optimization was accomplished by adjusting center wavelengths and bandwidths to minimize the size of intercorrection factors. Once optimized, the virtual filters were defined as follows: fat B sample, 3.508μm (2,851cm−1), and bandwidth of 0.032μm (26cm−1); fat B reference, 3.556μm (2812cm−1), and bandwidth of 0.030μm (24cm−1); lactose sample, 9.542μm (1,048cm−1), and bandwidth of 0.092μm (20cm−1); lactose reference, 7.734μm (1,293cm−1), and bandwidth of 0.084μm (14cm−1); protein sample, 6.489μm (1,541cm−1), and bandwidth of 0.085μm (20cm−1); protein reference, 6.707μm (1491cm−1), and bandwidth of 0.054μm (12cm−1); fat A sample, 5.721μm (1,748cm−1), and bandwidth of 0.052μm (16cm−1); fat A reference, 5.583μm (1,791cm−1), and bandwidth of 0.050μm (16cm−1). The bandwidth and its proximity to areas of intense water absorption had the largest effect on the intercorrection factors. The second objective was to quantify the impact of fatty acid chain length and unsaturation on fat B and fat A MIR measurements. Increasing the chain length increased the difference (i.e., MIR minus reference) between MIR prediction and reference chemistry by 0.0429% and by −0.0566% fat per unit of increase in carbon number per 1% change in fat, for fat B and fat A, respectively. Increasing the unsaturation decreased the difference (i.e., MIR minus reference) between MIR prediction of fat and chemistry for fat B by −0.4021% and increased fat A by 0.0291% fat per unit of increase in double bonds per 1% change in fat concentration.