• Application of Meta-Analysis and Machine Learning Methods to the Prediction of Methane Production from In Vitro Mixed Ruminal Micro-Organism Fermentation

      Ellis, Jennifer L.; Alaiz-Moretón, Héctor; Navarro-Villa, Alberto; McGeough, Emma J.; Purcell, Peter; Powell, Christopher D.; O’Kiely, Padraig; France, James; López, Secundino; Department of Agriculture, Food and the Marine; et al. (MDPI AG, 2020-04-21)
      In vitro gas production systems are utilized to screen feed ingredients for inclusion in ruminant diets. However, not all in vitro systems are set up to measure methane (CH4) production, nor do all publications report in vitro CH4. Therefore, the objective of this study was to develop models to predict in vitro CH4 production from total gas and volatile fatty acid (VFA) production data and to identify the major drivers of CH4 production in these systems. Meta-analysis and machine learning (ML) methodologies were applied to a database of 354 data points from 11 studies to predict CH4 production from total gas production, apparent DM digestibility (DMD), final pH, feed type (forage or concentrate), and acetate, propionate, butyrate and valerate production. Model evaluation was performed on an internal dataset of 107 data points. Meta-analysis results indicate that equations containing DMD, total VFA production, propionate, feed type and valerate resulted in best predictability of CH4 on the internal evaluation dataset. The ML models far exceeded the predictability achieved using meta-analysis, but further evaluation on an external database would be required to assess generalization ability on unrelated data. Between the ML methodologies assessed, artificial neural networks and support vector regression resulted in very similar predictability, but differed in fitting, as assessed by behaviour analysis. The models developed can be utilized to estimate CH4 emissions in vitro.
    • Capturing the economic benefit of Lolium perenne cultivar performance

      McEvoy, Mary; O'Donovan, Michael; Shalloo, Laurence; Department of Agriculture, Food and the Marine (Teagasc (Agriculture and Food Development Authority), Ireland, 2011)
      Economic values were calculated for grass traits of economic importance in Irish grass-based ruminant production systems. Traits considered were those that had the greatest potential to influence the profitability of a grazing system. These were: grass dry matter (DM) yield in spring, mid-season and autumn, grass quality (dry matter digestibility; DMD), 1st and 2nd cut silage DM yield and sward persistency. The Moorepark Dairy Systems Model was used to simulate a dairy farm. Economic values were calculated by simulating the effect of a unit change in the trait of interest while holding all other traits constant. The base scenario involved a fixed herd size and land area (40 ha), and an annual DM yield of 13 t/ha. The economic values generated under the base scenario were: € 0.152/kg for DM yield in spring, € 0.030/kg for DM yield in mid-season and € 0.103/kg for DM yield in autumn; € 0.001, € 0.008, € 0.010, € 0.009, € 0.008 and € 0.006 per 1 g/kg change in DMD for the months of April to September, respectively; € 0.03/kg for 1st cut silage DM yield, € 0.02/kg for 2nd cut silage DM yield; and − € 4.961 for a 1 percent reduction in persistency. Alternative scenarios were examined to determine the sensitivity of the economic values to changes in annual DM yield, sward utilisation and a scenario where silage production was the focus of the system. The economic values were used to calculate a total merit index for each of 20 perennial ryegrass cultivars based on production data from a 3 year plot study. The rank correlation between the merit index values for the cultivars under the base scenario and the scenario involving a reduction in herbage utilisation was 1.0, while that with the scenario involving reduced annual DM yield was 0.94. It is concluded that the total merit index can be used to identify cultivars that can generate the greatest economic contribution to a grass-based production system, regardless of system or intensity of grass production.
    • Comparison of herbage yield, nutritive value and ensilability traits of three ryegrass species evaluated for the Irish Recommended List

      Burns, G. A.; O'Kiely, Padraig; Grogan, D.; Watson, S.; Gilliland, T. J.; Department of Agriculture, Food and the Marine; RSF 07 526 (Teagasc (Agriculture and Food Development Authority), Ireland, 2016-01-13)
      This study examined 169 of the newest varieties of three ryegrass species, perennial (Lolium perenne L.), Italian (Lolium multiflorum Lam.) and hybrid (Lolium boucheanum Kunth), from Recommended List trials in Ireland. The traits examined were yield, dry matter concentration, three nutritive value traits (in vitro dry matter digestibility, water-soluble carbohydrate on a dry matter basis and crude protein concentration) and two ensilability traits (buffering capacity and water soluble carbohydrate concentration on an aqueous phase basis). Varietal monocultures of each species underwent a six cut combined simulated grazing and silage management in each of two years following sowing. Perennial ryegrass yielded less than both other species in one-year-old swards, but less than only Italian ryegrass in two-year-old swards, but generally had the higher in vitro dry matter digestibility and crude protein values. Italian ryegrass displayed the most favourable ensilability characteristics of the three species with perennial ryegrass less favourable and hybrid ryegrass intermediate. Overall, despite the high yields and favourable nutritive value and ensilability traits recorded, the general differences between the three ryegrass species studied were in line with industry expectations. These findings justify assessing the nutritive value and ensilability of ryegrass species, in addition to yield, to allow farmers select species that match farming enterprise requirements.
    • Concentrate feeding and feed ingredients for growing-finishing

      McGee, Mark; O’Riordan, Edward; Moloney, Aidan; Department of Agriculture, Food and the Marine (La Revue Scientifique, 2019-10-18)
      Small improvements in feed efficiency, especially during indoor ‘winter’ feeding periods, can have a relatively large influence on farm profitability. Increasing the level of concentrates in the diet reduces forage intake and increases live weight and carcass weight gains, although at a decreasing rate. Subsequent compensatory growth at pasture diminishes the advantage of concentrate supplementation of young cattle. High digestibility grass silage with moderate concentrate supplementation can sustain a large proportion of the cattle performance achieved on highconcentrate diets. Feeding management is more important when feeding concentrates ad libitum than as a supplement. The relative nutritive (and economic) value of by-product feed ingredients depends on their inclusion level in the ration, and the amount of concentrates fed.
    • Effect of different forage types on the volatile and sensory properties of bovine milk

      Faulkner, Hope; O'Callaghan, Tom; McAuliffe, Stephen; Hennessy, Deirdre; STANTON, CATHERINE; O'Sullivan, Maurice G.; Kerry, Joseph P.; Kilcawley, Kieran; Department of Agriculture, Food and the Marine; 13SN401 (Elsevier, 2017-12-08)
      The effect of 3 diets (grass, grass/clover, and total mixed ration) on the volatile and sensory properties of bovine milk was assessed over an entire lactation season. Little evidence was found of direct transfer of terpenes into raw milk from the different diets, and it is likely that the monocultures of ryegrass used with and without white clover were factors as these contained very few terpenes. Evidence of direct transfer of nonterpene volatiles from forage to the subsequent raw milks was probable; however, differences in the protein carbohydrate availability and digestion in the rumen appeared to have a greater contribution to volatile profiles. Pasteurization significantly altered the volatile profiles of all milks. A direct link between the milk fatty acid content, forage, and volatile products of lipid oxidation was also evident and differences in fatty acid content of milk due to forage may also have influenced the viscosity perception of milk. Irish sensory assessors preferred pasteurized milk produced from grass-fed cows, with least preference from milk produced from total mixed ration diets. β-Carotene content was significantly higher in milks derived from grass or grass/clover and appears to have directly influenced color perception. Toluene and p-cresol are both degradation products of β-carotene and along with β-carotene were identified as potential biomarkers for milk derived from pasture. The only correlation that appeared to influence the flavor of milk as determined using ranked descriptive analysis was p-cresol. P-Cresol appears to be responsible for the barnyard aroma of milk and is also likely derived from the deamination and decarboxylation of tryptophan and tyrosine due to the higher levels of available protein in the grass and grass/clover diets. The highest levels of p-cresol were in the grass/clover diets and are likely due to the degradation of the isoflavone formononetin in the rumen, which is present in white clover swards.
    • The effect of target postgrazing height on sward clover content, herbage yield, and dairy production from grass-white clover pasture

      Phelan, Paul; Casey, Imelda A.; Humphreys, James; Department of Agriculture, Food and the Marine; Teagasc Walsh Fellowship Programme; RSF 07-511 (Elsevier, 2013-01-16)
      White clover (Trifolium repens) is an important legume for grazed grassland that can increase the profitability and environmental sustainability of milk production. Previous experiments on mown grass-clover plots suggest that low postgrazing heights (PGH) can increase sward clover content and herbage production. However, this has not been tested in actual strip or rotational grazing systems with dairy cows. Furthermore, lowering PGH in grass-only swards (typically perennial ryegrass without white clover) has previously been associated with reduced milk yields per cow. The objective of this experiment was to investigate the effect of PGH by dairy cows on clover content, herbage production, and milk production from strip-grazed grass-white clover swards in Ireland. Three target PGH treatments of 4, 5, and 6 cm were in place for entire grazing seasons (February to November) for 3 consecutive years (2007 to 2009). Each treatment had a mean of 21 Holstein-Friesian dairy cows that strip-grazed a mean annual area of 10.2 ha. Postgrazing height was measured twice a day with a rising plate meter, and cows were moved to the next strip once the target PGH was reached. Annual fertilizer nitrogen input was 90 kg of N/ha for each treatment. The PGH treatment did not significantly affect annual milk yield (6,202 kg/cow), solids-corrected milk yield (6,148 kg/cow), fat, protein, or lactose yields (265, 222, and 289 kg/cow, respectively), cow liveweight (592 kg) or body condition score (3.01). The PGH treatment also had no significant effect on sward white clover content (196 g/kg). However, herbage production of both grass and clover were significantly higher with the 4-cm PGH treatment compared with the 6-cm treatment. Mean annual herbage yields were 11.1, 10.2, and 9.1 t of organic matter (OM)/ha for the 4-, 5-, and 6-cm PGH treatments, respectively. The lower herbage production in the 6-cm PGH treatment resulted in lower annual silage production, greater housing requirements, and a substantially higher net silage deficit (−1,917 kg of OM/cow) compared with the 5- or 4-cm treatments (−868 and −192 kg of OM/cow, respectively). Grazing to a PGH of 4 cm is therefore recommended for grass-white clover swards.
    • The effect of water-soluble carbohydrate concentration and type on in vitro rumen methane output of perennial ryegrass determined using a 24-hour batch-culture gas production technique

      Purcell, Peter J; Boland, T.M.; O'Kiely, Padraig; Department of Agriculture, Food and the Marine; RSF 07 517 (Teagasc (Agriculture and Food Development Authority), Ireland, 2014)
      The objective of this study was to examine the effects of water-soluble carbohydrate (WSC) concentration and type on the in vitro rumen methane (CH4) output of perennial ryegrass (PR) using a 24-hour batch-culture gas production technique. Dried and milled PR was incubated either alone (PR-O) or with added sucrose (PR-S), inulin (PR-I), or sucrose plus inulin (PR-S+I; sucrose:inulin ratio of 1:4) in sealed glass bottles [0.5 g total substrate dry matter (DM) per bottle] at 39 °C for 24 hours with buffered rumen fluid. The WSC types were added (except for PR-O) so that the WSC concentration in each fermentation bottle at the start of the incubation was either 180 (i.e., PR-O), 225, 270, 315, or 360 g/kg of total substrate DM incubated. There were linear decreases in CH4 output per gram of DM disappeared (CH4/ivDMD) and per mmol of total volatile fatty acid output (CH4/tVFA) with increasing WSC concentration in the incubated substrate. The WSC type had no effect on in vitro rumen CH4 output. It is concluded that since CH4/ivDMD and CH4/tVFA were reduced by increasing the concentration of WSC incubated with PR, it would be worthwhile to undertake in vivo experiments to examine these effects on in vivo CH4 emissions per unit of animal product.
    • Effects of fertiliser nitrogen rate to spring grass on apparent digestibility, nitrogen balance, ruminal fermentation and microbial nitrogen production in beef cattle and in vitro rumen fermentation and methane output

      O'Connor, Alan; Moloney, Aidan P; O'Kiely, Padraig; Boland, T. M.; McGee, Mark; Teagasc Walsh Fellowship Programme; Department of Agriculture, Food and the Marine; 11/S/105 (Elsevier, 2019-06-06)
      The effects of two fertiliser nitrogen (N) application rates - 15 (LN) or 80 (HN) kg N/ha - to Lolium perenne dominant swards in spring, on grass dry matter (DM) intake, digestion, rumen fermentation, microbial N production and N-balance in beef cattle, and in vitro fermentation and methane production were studied. Sixteen Charolais steers with a mean live weight (s.d.) of 475 (18.4) kg, were used in a completely randomised block design experiment and offered zero-grazed grass harvested 21-d post N application. The same grass was incubated in an eight-vessel RUSITEC in a completely randomised block design experiment. The HN treatment had a 540 kg/ha higher grass DM yield, and a 20 g/kg DM higher crude protein (CP) concentration compared to LN. There was no difference (P > 0.05) in DM intake, or in vivo DM, organic matter (OM) and N digestibility between treatments. Rumen fermentation variables pH, lactic acid, ammonia (NH3) and total volatile fatty acid (VFA) concentration were similar (P > 0.05) for both treatments. Nitrogen intake was 19 g/d higher (P < 0.05) for HN compared to LN. Total and urine N loss was 16 and 14 g/d greater (P < 0.05), respectively, for HN compared to LN, but faecal N loss did not differ (P > 0.05) between treatments. The quantity of N retained and N-use efficiency did not differ (P > 0.05) between LN and HN. Plasma urea concentration was 1 mmol/L greater (P < 0.05) for HN compared to LN. Estimated microbial N production was greater (P < 0.05) for HN compared to LN. In vitro NH3 concentrations were higher (P < 0.05) for HN compared to LN, whereas in vitro rumen pH, lactic acid and VFA concentrations and molar proportions did not differ (P > 0.05) between HN and LN. In vitro methane and total gas output were not different (P > 0.05) between treatments. Reducing fertiliser N application rate to grass in spring reduced total and urinary N excretion, which has environmental benefits, with no effects on in vitro methane output.
    • How much grassland biomass is available in Ireland in excess of livestock requirements?

      McEniry, Joseph; Crosson, Paul; Finneran, Eoghan; McGee, Mark; Keady, Tim; O'Kiely, Padraig; Department of Agriculture, Food and the Marine; RSF 07 557 (Teagasc (Agriculture and Food Development Authority), Ireland, 2013)
      Grassland is a dominant biomass resource in Ireland and underpins most animal production systems. However, other commercial uses for grassland biomass exist, including, for example, the production of biogas through anaerobic digestion for the generation of heat, electricity and transport fuel. The objective of this study was to estimate the annual grassland resource available in Ireland in excess of livestock requirements under six contrasting scenarios. Under current grassland management and production practices there is an estimated average annual grassland resource of ca. 1.7 million tonnes of dry matter (DM) available in excess of livestock requirements. Only a small proportion of this resource (0.39 million tonnes of DM per annum) would be available if the targets set out in ‘Food Harvest 2020’ were achieved. However, increasing nitrogen (N) fertiliser input (to the limit permitted by the E.U. Nitrates Directive) combined with increasing the grazed grass utilisation rate of cattle (from 0.60 to 0.80 kg DM ingested by livestock per kg DM grown) has the potential to significantly increase this average resource to 12.2 million t DM/annum, even when allowing for achievement of ‘Food Harvest 2020’ targets. Under these scenarios, alternative uses for grassland biomass such as anaerobic digestion and green biorefining would not compete with traditional dairy, beef and lamb production systems, but could provide an alternative enterprise and income to farmers.
    • Integration of high and low field 1H NMR to analyse the effects of bovine dietary regime on milk metabolomics and protein-bound moisture characterisation of the resulting mozzarella cheeses during ripening

      Boiani, Mattia; Sundekilde, Ulrik; Bateman, Lorraine M.; McCarthy, Daniel G.; Maguire, Anita R.; Gulati, Arunima; Guinee, Timothy P.; Fenelon, Mark; Hennessy, Deirdre; Fitzgerald, Richard J.; et al. (Elsevier, 2018-10-11)
      The influence of dairy cow feeding regime was investigated using 1H nuclear magnetic resonance (NMR). Two different NMR analytical systems were deployed: high field 1H NMR to investigate the influence on milk metabolomics and low field NMR to characterise proton relaxation linked to changes in the state of mozzarella cheese moisture during ripening. The metabolomics results showed that grass-based feeding increased the concentration of a biological marker that signifies near-organic milk production conditions. On the other hand, the investigation of cheese moisture distribution showed that grass-based diets reached final moisture partitioning in a shorter time, which implied the formation of a more compact protein structure in the cheese matrix. These results indicate that pasture-based dairying may be differentiated in terms of the provenance of milk produced along with the accrual of additional benefits during ripening of the resulting mozzarella cheeses.
    • Investigating the role of stocking rate and prolificacy potential on profitability of grass based sheep production systems

      Bohan, A.; Creighton, Philip; Boland, T.M.; Shalloo, Laurence; Earle, Elizabeth; McHugh, Noirin; Department of Agriculture, Food and the Marine; 15/S/696 (Elsevier BV, 2018-02-21)
      The objective of this study was to simulate and compare the profitability of a grass based sheep production system under three stocking rates and two prolificacy rates. Analysis was conducted using the Teagasc Lamb Production Model (TLPM), a stochastic budgetary simulation model of a sheep farm. Experimental data from the Teagasc Athenry Research Demonstration Flock was used to parameterise the model at three stocking rates (10, 12 and 14 ewes/ha) and two prolificacy potentials (1.5 and 1.8 lambs weaned per ewe joined to the ram). The TLPM assessed the performance of the key factors affecting profitability and was also used to evaluate the spread in profitability associated with some stochastic variables included in the analysis. The number of lambs weaned per hectare increased with stocking rate and prolificacy potential from 16 lambs/ha to 27 lambs/ha resulting in carcass weight produced per hectare ranging from 272 kg/ha to 474 kg/ha. Increasing stocking rates resulted in lower individual lamb performance from grass and milk, thereby increasing the proportion of lambs which required concentrate for finishing, which resulted in higher input costs on a per animal basis. As the number of lambs weaned per hectare increased, net profit increased from €361/ha to €802/ha. Across all stocking rates, increasing weaning rate from 1.5 to 1.8 lambs weaned per ewe joined increased net profit, on average, by €336/ha. Increasing stocking rate, at 1.5 lambs weaned per ewe joined, increased net profit on average by €15/ha while increasing stocking rate, at 1.8 lambs weaned per ewe joined increased net profit on average by €87/ha. Risk analysis showed that across all stocking rates the high prolificacy scenarios achieved greater profits across the variation in input variables. Results from this study indicate that lambs weaned per hectare linked with grass growth and utilisations are the key drivers of profitability on Irish grass based sheep production systems.
    • A note on the chemical composition and in vitro digestibility of contrasting stover components of maize grown in climatically marginal conditions and harvested at differing maturities.

      Lynch, J.P.; O'Kiely, Padraig; Doyle, Evelyn M.; Department of Agriculture, Food and the Marine; RSF 07 501 (Teagasc (Agriculture and Food Development Authority), Ireland, 2014)
      This study evaluated the nutritive value of three contrasting components of maize stover (leaf, upper stem, lower stem) at three harvest dates. The leaf component had a greater in vitro dry matter digestibility (DMD) and a lower NDF concentration, compared to the stem components. Delaying harvest reduced the in vitro DMD of the stem components to a greater extent than leaf, reflecting lower increases in the NDF and lignin concentrations in leaf tissue. The stem components of maize stover had a lower nutritive value than the leaf component, and had a larger decrease in digestibility with delayed harvest.
    • A note on the comparison of three near infrared reflectance spectroscopy calibration strategies for assessing herbage quality of ryegrass

      Burns, G. A.; O'Kiely, Padraig; Grogan, D.; Gilliland, T. J.; Department of Agriculture, Food and the Marine; RSF 07 526 (Teagasc (Agriculture and Food Development Authority), Ireland, 2014)
      Perennial ryegrass (n = 1,836), Italian ryegrass (n = 137) and hybrid ryegrass (n = 103) herbage was taken from harvested plots from the Irish national variety evaluation scheme and analysed for in vitro dry matter digestibility, water soluble carbohydrate concentration, crude protein concentration and buffering capacity. Spectral data were obtained using near infrared reflectance spectroscopy and three calibration strategies (global, species-specific or local) were utilised to relate the reference values to the spectral data. The local strategy generally provided the poorest estimation of herbage composition, with global and species-specific calibration strategies producing similarly accurate estimates of each quality trait. The higher accuracy and easier maintenance of the global strategy make it the recommended calibration method for analysing quality of ryegrass.
    • A note on the evaluation of the acid-insoluble ash technique as a method for determining apparent diet digestibility in beef cattle.

      McGeough, E.J.; O'Kiely, Padraig; Kenny, David A.; Department of Agriculture, Food and the Marine; RSF 05 224 (Teagasc, Oak Park, Carlow, Ireland, 2010)
      The objective was to determine if the acid-insoluble ash (AIA) method provided accurate estimates of in vivo apparent digestibility compared with the standard total faecal collection (TFC) method. Twelve steers, mean live weight 328 (s.d. 27.3) kg, were offered one of three diets based on whole-crop wheat (WCW) or a grass silage (GS) diet in a 4 × 4 latin square design. Apparent dietary digestibility was determined simultaneously using AIA and TFC methods. Agreement between the two methods depended on diet type, with acceptable agreement (a difference between the methods of 0.06), observed with the WCW-based diets. However, the strength of the agreement was weakened with the inclusion of GS. Agreement statistics were found to be a useful tool for assessing the relationship between the two methods of measurement.
    • A note on the fermentation characteristics of red clover silage in response to advancing stage of maturity in the primary growth

      King, Colman; McEniry, Joseph; O'Kiely, Padraig; Department of Agriculture, Food and the Marine; RSF 07 557 (Teagasc, 2012-12)
      This study investigated the silage fermentation characteristics of red clover (Trifolium pratense L., var. Merviot) harvested at five dates in the primary growth (at two week intervals from 12 May to 7 July). Despite the challenging herbage ensilability characteristics pre-ensiling [i.e. low dry matter (DM) concentration (142 to 178 g/kg), low water soluble carbohydrate concentration (51 to 118 g/kg DM) and high buffering capacity (552 to 639 mEq/kg DM], the silages preserved successfully and showed little evidence of clostridial activity (i.e. low concentration of butyric acid and ammonia-N). Stage of maturity at harvest had little effect on silage fermentation characteristics.
    • Outdoor grazing of dairy cows on pasture versus indoor feeding on total mixed ration: Effects on gross composition and mineral content of milk during lactation

      Gulati, Arunima; Galvin, Norann; Lewis, Eva; Hennessy, Deirdre; O'Donovan, Michael; McManus, Jennifer J.; Fenelon, Mark; Guinee, Timothy P.; Department of Agriculture, Food and the Marine; Dairy Levy Trust Co-Operative Society Limited; et al. (Elsevier, 2017-08-15)
      The influence of feeding system and lactation period on the gross composition, macroelements (Ca, P, Mg, and Na), and trace elements (Zn, Fe, Cu, Mo, Mn, Se, and Co) of bovine milk was investigated. The feeding systems included outdoor grazing on perennial ryegrass pasture (GRO), outdoor grazing on perennial ryegrass and white clover pasture (GRC), and indoors offered total mixed ration (TMR). Sixty spring-calving Holstein Friesian dairy cows were assigned to 3 herds, each consisting of 20 cows, and balanced with respect to parity, calving date, and pre-experimental milk yield and milk solids yield. The herds were allocated to 1 of the 3 feeding systems from February to November. Milk samples were collected on 10 occasions over the period June 17 to November 26, at 2 or 3 weekly intervals, when cows were on average 119 to 281 d in lactation (DIL). The total lactation period was arbitrarily sub-divided into 2 lactation periods based on DIL, namely mid lactation, June 17 to September 9 when cows were 119 to 203 DIL; and late lactation, September 22 to November 26 when cows were 216 to 281 DIL. With the exception of Mg, Na, Fe, Mo, and Co, all other variables were affected by feeding system. The GRO milk had the highest mean concentrations of total solids, total protein, casein, Ca, and P. The TMR milk had the highest concentrations of lactose, Cu, and Se, and lowest level of total protein. The GRC milk had levels of lactose, Zn, and Cu similar to those of GRO milk, and concentrations of TS, Ca, and P similar to those of TMR milk. Lactation period affected all variables, apart from the concentrations of Fe, Cu, Mn, and Se. On average, the proportion (%) of total Ca, P, Zn, Mn, or Se that sedimented with the casein on high-speed ultracentrifugation at 100,000 × g was ≥60%, whereas that of Na, Mg, or Mo was ≤45% total. The results demonstrate how the gross composition and elemental composition of milk can be affected by different feeding systems.
    • Selection of calibration sub-sets to predict ryegrass quality using principle component analysis for near infrared spectroscopy

      Burns, G. A.; O'Kiely, Padraig; Gilliland, T. J.; Department of Agriculture, Food and the Marine; RSF –07 526 (British Grassland Society, 2015-09)
      Near infrared reflectance spectroscopy (NIRS) has become the routine method of assessing forage quality on grass evaluation and breeding programmes. NIRS requires predictive calibration models that relate spectral data to reference values developed using a calibration set (Burns et al. 2013). The samples that form the calibration set influence the accuracy and reliability of these models and need to be representative of samples that will likely be analysed (Shenk and Westerhaus, 1991; Burns et al. 2014). Analysing samples from the calibration set using reference techniques has a significant cost and time associated and needs to be considered in the context of the desired accuracy and robustness of calibration models. Calibration selection techniques can therefore maximise the accuracy and robustness of calibration models whilst reducing the number of samples requiring reference analysis. One such method is principal component analysis (PCA; Shenk and Westerhaus, 1991) whereby Shetty et al. (2012) reported that the number of samples could be reduced by up to 80% with a minimal loss in accuracy of calibration model. PCA selects representative calibration sub-sets through plotting all the samples in hyper-dimensional space, based on spectral data, and a sample is selected to represent a local neighbourhood cluster of samples for reference analysis. The aim of this research was to assess the accuracy of NIRS calibration models for buffering capacity, in vitro dry matter digestibility (DMD) and water soluble carbohydrate (WSC) content developed using calibration sub-sets selected by PCA.