Rootstock-scion genotype and environment interaction in a South African citrus breeding programme

[摘要] English: Cultivars with high and stable genetic potential for production and quality is the main goal of a fruit-breeding programme and are assessed in multi-environment trials due to the influence of rootstocks as well as climate on yield and quality in citrus. In a crop where grafting is essential, the manifestation of the scion's genotype is dependent on the rootstock on which it is grafted as well as the environment in which the scion-rootstock combination (stion) is grown. The problem with grafted trees is that part of what is measured in these trials is the rootstock's reaction to the environment as well as the rootstock's interaction with the scion and vice versa, which constitutes complex genotype x environment interactions (GEI). The aim of this study was to successfully separate the Genotype (G) and GEI of the stion into a scion and a rootstock G and GEI.Data used in this investigation emanated from Phase II trials within the South African citrus breeding programme and comprised of five citrus scion types namely grapefruit (Citrus paradise), midseason oranges (C. sinensis), Valencia oranges (C. sinensis) early mandarins (C. reticulate) and late mandarins (Citrus paradise). Rootstock selections that were included were: Van Stadens Rough lemon (C. jambhiri, Tenaka), Volckamer lemon (C. volkameriana V. Ten. & Pasq.), Empress Rosehaugh (C. reticulate Swingle) and Carrizo citrange (Poncirus trifoliate x C. sinensis). Three localities were included namely Messina, Malalane and Friedenheim for the univariate study. Data from Malalane over five years was used to test two multivariate models namely AMMI and GGE.The multivariate analysis confirmed citrus scion types as mega environments in relation to rootstocks. No single mega environment for rootstock selection, that was both discriminative and stable with regard to all the traits in this trial, could be found. Mega environments were trait specific with, for instance grapefruit, representing the ideal test environment for peel thickness and Valencia the most stable environment with regard to TSS:TA ratio. It was also found that the AMMI model was able to separate and quantify the contribution of the scion and rootstock to the stion in a single physical environment (i.e. same climate, same soil, same production practices). As was expected, the scion contribution was found to be more prominent than the rootstock contribution for most of the traits. There were exceptions such as in grapefruit, where the rootstocks were responsible for 53.21% of the yield effect as opposed to the 27.50% contribution of the scion and 54.96% with regard to peel thickness opposed to the 38.23% of the scion. GEI regarding scion (G) x rootstock (E) was significant but not for rootstock (G) x scion (E). This implies that the rootstocks in some or another way influenced the scions whereas the scions had no significant influence on the rootstocks.With a good insight into the biplot theory, the interpretation of the visual aspects of both the AMMI and GGE were found to be easy and beneficial. A dataset can generate a multitude of graphs which can render information at a quick glance but still with scientific context.The main consequence of applying multivariate models is the simplified and thus cost effective trial layout that it facilitates. There is no need any more for separate rootstock and scion trials. These trials can now be combined, incorporating selections from both the scions and rootstock programmes, saving orchard space and time but generating more information. The envisaged outcome of the study is a statistical method that in future trials would enable breeders to recommend the best rootstock-scion combination from time and cost effective evaluation of promising selections from the South African citrus breeding programme.

[发布机构] University of the Free State