Modelling and prediction of dry matter yield of perennial ryegrass cultivars sown in multi-environment multi-harvest trials in south-eastern Australia

Highlights

• Accurate seasonal predictions of ryegrass cultivar DM yields using LMMs.

• Base AR 37 and Bealy NEA2 were the best performing cultivars based on annual ranking.

• Seasonal variability was found to be larger than genotypic variability.

• An antedependence of order 3 was the best residual covariance structure for harvests.

Abstract

With over 60 commercial perennial ryegrass cultivars available on the market in Australia, selecting cultivars with high dry matter (DM) yield and economic profitability requires accurate estimates of their DM yield across target environments. This study, using data from multi-environment multi-harvest (MEMH) trials conducted in south-eastern Australia, derived accurate seasonal predictions of DM yield of these cultivars using linear mixed models (LMM). Base AR37 and Bealy NEA2 were found to be the best performing cultivars in most of the seasons in the target south-eastern Australian environments. Seasonal variability was found to be larger than genotypic variability as usually is the case in multi-environment trials. We have provided details of the LMM methodology used, along with ASReml-R code, to enable others to apply it in similar studies with appropriate changes as required for dataset used. The statistical theory underlying this methodology has also been briefly described in an Appendix for interested readers.

Introduction

Perennial ryegrass (Lolium perenne L.) forms the basis of “home grown” forage supply on dairy farms in south-eastern Australia. With over 60 cultivars commercially available on the market, selecting the most productive (in terms of dry matter (DM) yield) and economically profitable cultivars requires accurate estimates of DM yield performance of these cultivars based on data from multi-environment multi-harvest (MEMH) trials. These estimates are also used to develop seasonal forage value indices (FVI) that combine DM yield performance of cultivars with their relative economic values. These indices assist farmers to select better performing cultivars with the best possible balance of attributes for their production system and environment in different seasons of the year. Seasonal FVIs have been derived for perennial ryegrass in Ireland (McEvoy et al., 2011), New Zealand (Chapman et al., 2017) and Australia (Leddin et al., 2018).

To enable reliable FVI-based selection of better performing cultivars, this study aimed to derive accurate seasonal predictions of DM yield performance of commercial perennial ryegrass cultivars in the target population of south-eastern Australian environments using data from MEMH trials. The predictions were derived using a linear mixed model (LMM) methodology as appropriate for the analysis of data from MEMH trials in perennial crops. This paper provides a more detailed version of this methodology, along with ASReml-R code, to enable others to implement it in similar studies. While the statistical methods used are well established, this is the first report on an application of these statistical methods for perennial ryegrass cultivar evaluation using data from MEMH trials. The statistical theory underlying this methodology has been briefly described in the Appendix.