Two-stage method linking spatial point processes and linear mixed models for plant root tip data

Neale, Luke (2022) Two-stage method linking spatial point processes and linear mixed models for plant root tip data. Honours thesis, University of Southern Queensland. (Unpublished)

Text (Thesis)
Available under License Creative Commons Attribution Non-commercial Share Alike 4.0.

Download (2MB) | Preview


Spatial point processes allow for model fitting to spatial point pattern data. These spatial models estimate parameters, including covariates, based on the spatial location of points of interest. Despite the effectiveness of this methodology in analysing point pattern data, the software has difficulty in terms of computation for large, replicated experiments. In these circumstances, the inferences that can be made from spatial point processes are quite limiting and this promotes the use of a two-stage method. The presence of design parameters from a replicated experiment and research interest in relationships between genotypes advocates the implementation of the linear mixed model framework in the second stage of the analysis. The viability of such a two-stage method is investigated in this study with a weighted and unweighted use of the linear mixed model compared. The linking of spatial point processes and linear mixed models has not been investigated before, and the findings will therefore shape similar analyses in the future.

The development of spatial point pattern data can occur through the use of aeroponic platforms. In this study, the spatial point pattern data represents the position of plant root tips. This data was collected from a designed experiment on wheat plants grown in aeroponic platforms conducted at the Catholic University of Louvain in Belgium. The hidden nature of plant root traits make their measurement difficult and various methods have been implemented in an attempt to increase the accuracy in measuring below ground traits. Aeroponic platforms allow accessibility to plant root systems in a non-destructive manner, aiding the measurement of below ground traits. Often, the primary research aim in these trials is to determine variation between genotypes. A range of below ground traits can be investigated when attempting to determine genotypic variation in the root architecture of plants. Such traits can be measured and investigated using the spatial locations of plant root tips.

This study showed that linking spatial point processes and linear mixed models through the use of a two-stage method is viable. The estimated parameters from the spatial point processes were used as the response variable in the linear mixed models, where differences in the results were evident. The unweighted linear mixed models showed difficulty in estimating genetic variance across the width of the plant root systems. This was not the case for the weighted linear mixed models, which accounted for the uncertainty in the parameter estimates to allow genetic variance to be estimated. This study also highlighted key components of this two-stage method which could be improved in the future.

Statistics for USQ ePrint 51740
Statistics for this ePrint Item
Item Type: Thesis (Non-Research) (Honours)
Item Status: Live Archive
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Sciences (6 Sep 2019 - 31 Dec 2021)
Supervisors: King, Rachel; Kelly, Alison
Qualification: Bachelor of Science (Honours)
Date Deposited: 28 Jun 2023 22:46
Last Modified: 28 Jun 2023 22:46
Uncontrolled Keywords: plant root; spatial point; data
Fields of Research (2020): 49 MATHEMATICAL SCIENCES > 4905 Statistics > 490501 Applied statistics

Actions (login required)

View Item Archive Repository Staff Only