Hamilton Institute Seminar

Wednesday, November 23, 2022 - 13:00 to 14:00
Hamilton Institute Seminar room (317), 3rd floor Eolas Building, North Campus, Maynooth University

Speaker: Professor Caroline Brophy, Trinity College Dublin

Title: "Statistical modelling of multi-site biodiversity experiments using random coefficient models"

Abstract: 
If an experiment is repeated at multiple sites, the inference and generalisability of the results from the experiment can be enhanced compared to a single-site study. For example, if multiple sites are selected at random across a country and a common experiment implemented, the results can generalise to the scale of the country, whereas the results from a single-site study may be confounded with the conditions of the selected site. There are several ways to analyse data from multi-site experiments, for example, including site as a random effect in a linear mixed modelling framework would account for the hierarchical nature of the data. However, the effects from experimentally manipulated treatments could vary from site to site and in this case, random coefficient models, where individual model parameters are assumed to vary from site to site, may be more suitable.

In recent years, results from studies of agronomic grassland ecosystems have shown that increasing the species diversity of an agronomic grassland can improve outcomes for the system, such as yield and resistance to weed invasion. With careful selection of species, this strategy can also increase the sustainability of agronomic grasslands by reducing the requirements for synthetic nitrogen fertiliser and thus alleviating the negative environmental consequences of synthetic fertilisers.

In this talk, random coefficients models for analysing multi-site biodiversity experiments will be introduced. The methods will be applied to a multi-site grassland biodiversity experiment that was carried out across five sites in Norway. At each site, single-species monocultures, and multi-species mixtures varying in the number and proportions of up to seven agronomic species, were sown at low and high nitrogen fertiliser and measured over a period of three years. The application of the statistical modelling approach to larger scale biodiversity studies will also be discussed.

Bio:
Caroline Brophy is a Professor in Statistics at Trinity College Dublin, Ireland. She has a BSc in Mathematics and Statistics and a PhD in Statistics, both from University College Dublin, Ireland. Prior to moving to Trinity College Dublin in 2019, she was an Assistant Professor at Maynooth University in the Department of Mathematics and Statistics since 2008. Her research programme is primarily in the area of developing statistical models for applications in ecology, agronomy and environmental studies and current work is funded by SFI, DAFM and DAERA, Teagasc and the Norwegian Research Council. She is also lead coordinator of the upcoming Marie Skłodowska-Curie Doctoral Network titled ‘LegumeLegacy - Optimising multiple benefits of grass, legume and herb mixtures in crop rotations: modelling mechanisms and legacy effects’, bringing together 15 partner institutions across Europe.

Virtual Participation: Join Zoom

Meeting ID: 880 1818 5733
Passcode: 279596