NMMU logo
Movement Ecology Workshop 2015

Part B: Data Analysis III

Back to roadmap home

10. RELATIONSHIPS BETWEEN LOCATIONS AND ENVIRONMENTAL DATA

This was not something covered specifically in the workshop, but we compiled some thoughts on the subject. Before we start, here are some very strong cautions and recommendations that came up:

  1. If you are looking to extract some environmental data, speak to an oceanographer (who is working in your study region). Some of the satellite-derived products are better than others, some have undergone more processing than others, and as ecologists we often cannot discriminate among these nuances.

  2. First start with a hypothesis and then look for data to test it; don't simply test a bunch of datasets because those are the only ones you could find online.

  3. Be aware of mismatches in space and time between environmental data and track locations.

  4. Be aware of time lags in ecological processes.

  5. Manage your workflow: environmental datasets can be treated in the same way as tracking data in terms of naming conventions, metadata files, etc.

 

In these kinds of analyses, the objective is generally to look for drivers of animal behaviour, habitat use and/or movement.

The approach would be to integrate/extract environmental data from either track locations or inside home ranges. Possible tools include the extract function in the R package raster, or extract by multipoint (and similar functions) in ArcGIS.

Possible data sources (make sure you select the "science quality data" where appropriate) include:

A suggestion: use netCDF (.nc) files - they are a good way to store very large datasets, otherwise .csv files are easy to handle. For information on how to access and manipulate netCDF files in R, read the PDF associated with the R package 'ncdf4'. Panoply (NASA) is also neat programme to visualise and handle .nc files.

 

Suggestions for analyses included:

  • GLMMs (but these don't take space and time into account)
  • Resource selection/utilization functions
  • nLME models
  • GAMs
  • Geostatistics
  • Spatial-temporal models
  • Other multivariate statistics
  • Pairs plot in R (for first-step, simple correlations among variables)

Back to roadmap home

 

Nelson Mandela Metropolitan University

Telephone: 041-504 1111 | Fax: 041-504 2574 / 2731

E-mail: info@nmmu.ac.za

PO Box 77000 | Nelson Mandela Metropolitan University

Contact usPort Elizabeth | 6031 | South Africa

YouTube Twitter Facebook Connect@nmmu

 

© NMMU 2005-2014

A-Z Index

NMMU site map