Thesis subject

Agent-based modelling of herd size dynamics – assessing the determinants of herd sizes in Africa’s ungulate populations

Nature conservation organisations need to know population sizes and trends. They have to assess population sizes for this. However, population sizes are always estimated based on extrapolations of counts, and these estimates are highly sensitive to herd size. A single large herd in your sample may lead to severe over-estimations of population sizes, and vice versa missing the large herd in your sample may lead to severe under-estimation of population sizes. Especially when the herd is missing in the sample, it is worthwhile to know whether it may have been present. Therefore we need more contextual information to understand whether you could have found these large herds, given what we know about the species and environmental conditions.

This project is part of the larger project ‘Digital Twins of Nature’.

Short description
Monitoring animal populations is one of the core activities of any nature conservation organisation, as it gives valuable insights in population sizes and trends. However, nature areas are often very large and poorly accessible, making reliable counts particularly challenging as we cannot be everywhere all the time. Even aerial surveys typically do not cover 100% of the area. The result is that population sizes are always estimated based on extrapolations of counts, and these estimates (and associated uncertainties) are highly sensitive to herd size. A single large herd in your sample may lead to severe over-estimations of population sizes, and vice versa missing the large herd in your sample may lead to severe under-estimation of population sizes. Especially when the herd is missing in the sample, it is worthwhile to know whether it may have been present. Therefore we need more contextual information to understand whether you could have found these large herds, given what we know about the species and environmental conditions.

Being part of a larger project called ‘Digital Twins of Nature’, this sub-project is aimed at better understanding the dynamics of group sizes in Africa’s large ungulate populations, which are known to distort population counts significantly. In this project you will develop an agent-based model that will help explore under which conditions (e.g., food abundance/scarcity and predation) animals form large herds and split up again.

Objectives

  1. Understand from literature and empirical data the most important behavioural and context variables that explain herd sizes in African ungulates
  2. Build an Agent-Based Model expressing the literature findings and usable for comparing counts of animals with actual population sizes
  3. Perform sensitivity analysis and validation on the model
  4. Validate the model against field data

Tasks
The work in this master thesis entails:

  • To collect articles on herd size in ungulates, but also on ABMs of animal observations, and understand the field of research
  • To familiarize yourself with and use the open access Netlogo software (NetLogo Home Page (northwestern.edu)), which includes a library of example models
  • To use the information provided above and plan, program and run the agent-based model with different configurations to test the hypotheses, provide sensitivity analyses and validate the model with empirical data
  • To report in detail and assist the supervisors with the writing of an article suitable for publication after the study is completed.

Literature

  • Arts, K., van der Wal, R., & Adams, W. M. (2015). Digital technology and the conservation of nature. Ambio, 44(4), 661-673.
  • Khaemba, W. M., Stein, A., Rasch, D., De Leeuw, J., & Georgiadis, N. (2001). Empirically simulated study to compare and validate sampling methods used in aerial surveys of wildlife populations. African Journal of Ecology, 39(4), 374-382.
  • Western, D., & Lindsay, W. K. (1984). Seasonal herd dynamics of a savanna elephant population. African Journal of Ecology, 22(4), 229-244.

Requirements

  • Recommended courses: Programming in Python (INF-22306), Agent-Based Modelling of complex adaptive systems (INF-34806)
  • Required skills/knowledge: Basic programming skills and interest in animal behaviour

Key words: Agent-based modelling, herd size, computational socio-ecological simulation

Contact persons

  • gertjanhofstede@wur.nl
  • koen.dekoning@wur.nl