On 7 February 2018, the RSS Merseyside Local Group hosted an afternoon session on the topic of 'Wildlife Statistics'. There were 27 people in attendance, approximately half of whom were RSS members. As usual, the majority of attendees were from a range of departments at the University of Liverpool. However, there were also attendees from the Health and Safety Executive, Durham University, and the Animal and Plant Health Agency.
Dr Marcus Rowcliffe (of the Zoological Society of London and Imperial College London) gave the first talk, titled 'Theory and practice in estimating abundance of unmarked wildlife using camera traps'. Camera traps are cameras fitted with infra-red sensors that trigger pictures to be taken when an animal moves within the sensors range. When using camera traps to estimate abundance of animal populations various issues exist, such as the reality that camera traps don’t have 360° detection, meaning that the direction the animal approaches the trap is important.
Initially, a random encounter model (such as those employed to model gas molecules) was employed. However, many real world variables such as vegetation, weather or topography could affect the animal’s behaviour. Consequently an integrated maximum likelihood approach is being developed, which collapses such variables into single measures to include in the model. For example, animal activity level and animal speed were collapsed to give movement rate. It was noted that non-random movement of animals was acceptable, given that the cameras themselves were placed randomly with respect to animal density.
The methods allowed multi-species models to be fitted to allow strength to be borrowed across different parameters and species. It was important to correct for confounding factors (eg it is easier to detect a puma than a mouse in a camera trap). In addition, consideration had to be taken when placing cameras in certain environments. For example, when recording lion abundance, lions were expected to be nocturnal but a large number were recorded during the day. Cameras had been placed on trees (there were few other options to attach cameras to), which provided shade. The lions slept in the shade during the day, triggering the traps and causing oversampling.
Overall, it was highlighted that multi-species modelling approaches allowed strength to be borrowed across variables and species, allowing better modelling of animals seldom detected in camera traps.
After a break for tea and coffee, the session resumed with a talk from Richard Glennie (of the University of St Andrews) titled 'Encounters between animals and detectors in wildlife abundance surveys'. The aim of the work was to estimate the number of animals observed versus the number of animals missed. In wildlife surveys, distance sampling assumes animals remain stationary during the survey, providing a 'snapshot in time'. However from movement ecology, telemetry can be obtained by tagging animals, and expertise about the behaviour of the animals can be sought. Encounter modelling links movement ecology and distance sampling, employing methods from time-to-event analyses.
If the path of the animal and the time of the recorded observation is known, detection parameters can be predicted. In real life, the path of the animal is not known, so a model is needed to predict the path and average over all the possible paths between encounters with same animal. This would require integration over infinitely many paths. Instead a fine grid is used, and hidden Markov Models are employed to calculate across all the paths through the grid.
Examples of applications of the method were shown for dolphin populations and jaguar populations. It was noted in the jaguar example, that the relationship between individuals could be seen – there was around 75% overlap in territories between neighbouring jaguars.
The next meeting of the RSS Merseyside local group will take place on Tuesday 5 June, and is titled Predictions within Sport, with confirmed speakers Professor Ian McHale (of the University of Liverpool), and Tom Hughes (Manchester United). Further details of this, and all future talks, are available on the Merseyside local group website: https://sites.google.com/site/rssmerseyside/.