Skip to content

Latest commit

 

History

History
56 lines (35 loc) · 3.64 KB

README.md

File metadata and controls

56 lines (35 loc) · 3.64 KB

Climate Effects on Pear Production: A Multivariate Analysis

Project for Multivariate and Hierarchical Data (3565) | Master of Statistics, Hasselt University

Team: Kate Igharas, Sara Man, Matteo Venturini, Shanbo Zhao

Full report

For a complete overview of the experimental design, statistical models, and in-depth discussion, please see the full project report.

Project Summary

This project investigates the impact of four projected 2050 climate scenarios on the quality, growth, and soil conditions for two pear varieties: Conference and Doyenne du Comice. Using data from a controlled ecotron experiment, we applied a range of multivariate and hierarchical statistical techniques to analyze the complex relationships between climate, soil, and fruit development.

The core research question was: To what extent do different climate scenarios affect the quality and growth of pears, and how do these effects differ between species?

Methodology Highlights

To answer our research question, we employed a robust analytical framework designed to handle the hierarchical and longitudinal nature of the data. The key methodologies used were:

  1. Linear Discriminant Analysis (LDA): To explore how different climate scenarios influenced soil properties and to identify the key soil variables that drive separation between the scenarios.
  2. Linear Mixed-Effects Models (LMM):
    • To analyze the continuous quality score (0-100) of pears, accounting for the nested structure of the data (pears are nested within trees).
    • To model the log-transformed pear size over a 24-week period, capturing individual growth trajectories while accounting for repeated measurements.
  3. Generalized Estimating Equations (GEE): To analyze the binary quality outcome (good vs. poor quality) and understand the population-average effects of climate and species on the probability of producing good quality fruit.

Key Findings & Visualizations

Our analysis revealed significant species- and scenario-specific effects on both pear quality and growth.

1. Conference Pears Consistently Outperform Doyenne Pears

Across all climate scenarios, the Conference variety showed higher quality scores and grew larger than the Doyenne variety.

Figure 3: Bar plot of binary quality index by pear variety and climate scenario Binary Quality Index

2. A Trade-Off Between Fruit Quality and Size

Scenario 2 (focused on active CO₂ removal) produced the highest quality fruit. However, it also resulted in the smallest fruit size. This suggests a potential physiological trade-off between quality and size under climate intervention.

Figure 4: Mean pear size over time by species and climate scenario Mean Pear Size Over Time

3. Climate Scenarios Drastically Alter Soil Properties

The LDA revealed that each climate scenario creates a distinct and highly separable soil profile. Scenarios 1 and 4 (worst-case and transportation-focused) were characterized by lower-than-average nutrient levels, while Scenarios 2 and 3 (CO₂ removal and sustainable energy) showed higher levels.

Figure 6: Linear Discriminant Analysis score plot LDA Score Plot

How to Run This Project

  1. Clone the repository: git clone https://github.com/maven2306/pear-climate-analysis.git
  2. Open the project in RStudio.
  3. The R packages required for this analysis are listed in the scripts, including lme4, geepack, and dplyr.
  4. Run the scripts in the /code folder.