Back to the roadmap home Next to Part B: Data Analysis II
The next three sections are deliberately not numbered because they are not necessarily sequential, and each analysis can inform the others. For example, you could use movement paramaters to inform segmentation, or compare the movement parameters among different segments. (Just be aware of circular arguments...)
> SEGMENTATION
The objective of segmentation is to split the tracks into homogeneous portions. The signals that could be used to inform the segmentation include:
-
speed
-
turning angles
-
distance
-
first-passage time
-
residence time... etc.
The tools available in R include:
-
change point detection
-
cghseg package
-
adehabitatLT: lavielle function
-
clustering
-
mixtools: normalmixEM
-
Rmixmod: mixmodcluster
-
kmeans
Or you could cut/segment the data in MS Excel.
> MOVEMENT PARAMETERS
One objective of quantifying the movement parameters could be in inform the signals for segmentation. Alternatively, as mentioned above, it could be to compare movement parameters among different states, or among different individuals.
Movement paramaters can be determined in the adehabitatLT package by:
-
ltraj summaries
-
first-passage time
-
residence time
They may also be measured as ancilliary data using other devices (like accelerometers).
> STATE IDENTIFICATION
The objectives could be to:
-
Reconstruct continuous state sequences
-
Identify discrete states that are either
a. known a priori
b. unknown a priori
Note: Objective 2a may seem strange; why identify states when you know them... But the point here may be, for example, to train and test a model with tracks of known states (from observers, cameras, or other ancilliary sensors) that can be applied to tracks of unknown states to increase sample size for understanding behaviour with reference to environmental data. Some examples of these kinds of analyses include:
Joo 2013 A behavioural ecology of fishermen: hidden stories from trajectory data in the Northern Humboldt Current System. PhD Thesis.
Bestley et al. 2012 Integrative modelling of animal movement: incorporating in situ habitat and behavioural information for a migratory marine predator
Hanks et al. 2011 Velocity-Based Movement Modeling for Individual and Population Level Inference
Back to the roadmap home Next to Part B: Data Analysis II