高寒草原天然牧草营养品质近红外光谱预测模型的建立

Prediction Model Establishment of Near Infrared Reflectance Spectroscopy for Nutrient Quality of Natural Forage in Alpine Grassland

本研究旨在探讨近红外光谱(NIRS)技术在定量分析高寒草原天然牧草营养品质的可行性。试验于2015—2018年,每年6—9月份,从青藏高原高寒草原放牧研究样地收集不同放牧强度(0、3.6、5.3、7.6只/hm2)的混合和4个优势种(莎草科矮生嵩草、蓼科珠芽蓼、蔷薇科金露梅和豆科锦鸡儿)牧草样品共计1280份,随机分为定标样品集(n=854)和预测样品集(n=426),建立天然牧草干物质(DM)、粗脂肪(EE)、粗蛋白质(CP)、酸性洗涤纤维(ADF)、中性洗涤纤维(NDF)含量及体外干物质消化率(IVDMD)和估算代谢能(EME)营养品质近红外光谱(NIRS)预测模型,实现天然牧草营养品质的快速和准确评估。结果显示:牧草DM、CP、EE、ADF、NDF含量及IVDMD和ME变异较大;DM、CP、ADF、NDF含量及IVDMD和EME的NIRS预测模型的验证决定系数(R2 CV)为0.992~0.999,外部验证相对分析误差(RPD)为3.82~5.97,取得了最佳的定标效果,且定标方程均具较好的预测能力,能够成功应用于日常分析。EE含量的NIRS预测模型的R2 CV为0.837,外部RPD为2.30,定标效果不理想,定标模型虽不能用于准确的定量分析测定,但仍能应用于牧草EE含量的粗略测定。综上所述,本研究建立的高寒草原天然牧草DM、CP、ADF、NDF含量及IVDMD和EME的NIRS预测模型能够成功应用于日常分析。

The objective of this study was to explore the feasibility of near infrared reflectance spectroscopy(NIRS)technology in quantitative analysis of the nutrient quality of natural forage in alpine grassland.During 2015 to 2018 years,from June to September of each year,the mixed and four dominant species(Kobresia hu⁃milis,Polygonum viviparum,Potentilla fruticosa and Caragana sinica)with different grazing intensities(0,3.6,5.3 and 7.6 sheep/hm2)were collected from the alpine grassland of Qinghai Tibet plateau.A total of 1280 forage samples were randomly divided into calibration sample set(n=854)and prediction sample set(n=426).The NIRS prediction models of forage nutritional quality were established,including the contents of dry matter(DM),ether extract(EE),crude protein(CP),acid detergent fiber(ADF)and neutral detergent fiber(NDF)and in vitro dry matter digestibility(IVDMD)and estimated metabolizable energy(EME),to realize the rapid and accurate evaluation of the nutrient quality of natural forage.The results showed that the contents of DM,CP,EE,ADF,NDF and IVDMD and EME of forage varied greatly;the determination coef⁃ficients of cross⁃validation(R2 CV)of contents of DM,CP,ADF,NDF and IVDMD and EME predicted by NIRS prediction model was range from 0.992 to 0.999,the external ratio of performance to deviation for vali⁃dation(RPD)was range from 3.82 to 5.97,it got the best calibration result,the calibration equations had good prediction ability and can be successfully applied to daily analysis.The R2 CV of EE content predicted by NIRS prediction model was 0.837,and the external RPD was 2.30,the calibration effect were not ideal.Al⁃though the calibration model could not be used for accurate quantitative analysis,it could still be applied to rough determination of EE content of forage.In conclusion,the established NIRS prediction models of contents of DM,CP,ADF,NDF and IVDMD and EME in this study can be successfully used in daily analysis.