Seasonality in daily movement patterns of mandrills revealed by combining direct tracking and camera traps

This dataset, containing three CSV files and one R code, was collected in a field study on wild mandrill (Mandrillus sphinx) groups in Moukalaba-Doudou National Park, Gabon. We analysed the data to examine the temporal variations in movement rates of mandrill groups and their seasonal changes.

Summary of Methods and Results

In a 400-km2 rainforest area in Moukalaba-Doudou National Park, Gabon, we tracked unidentified groups 46 times in 2009–2013; in the same area, we systematically placed 157 terrestrial camera traps in 2012–2014 and recorded groups 309 times. Generalised additive mixed models (GAMM) of the tracking data (GPSerror.csv, TrackData.csv) indicated that the group travel speed varied with time and season: In the fruiting season, the movement rate fluctuated with time in a bimodal pattern; in the non-fruiting season, it increased monotonously with time. Predicted day ranges were longer in the fruiting season (7.10 km) than in the non-fruiting season (6.18 km). These seasonal differences suggest responses to changes in food resources and temperature. Camera-trap detection rates (CameraData.csv) showed similar temporal and seasonal patterns to the tracking data, allowing us to generalise our findings to the population level. Moreover, cameras never detected mandrills at night, and we observed that they slept high in trees and hardly moved until the next morning, all suggesting their strict avoidance of night-time movement.

Methods
GPSerror.csv (data on the measurement error of a handy GPS device) was collected from a handy GPS device (Garmin GPSmap 62s) left in a particular place in the forested study area. We used a generalised linear model (GLM) to analyse the distribution of the measurement error of GPS coordinates.

TrackData.csv (data on direct tracking of mandrill groups) was collected by the direct tracking of mandrill groups in 2009–2013 using handy GPS devices (Garmin GPSmap 60csx/62s). We used generalised additive mixed models (GAMMs, with the R package "gamm4") to examine the temporal and seasonal changes in movement rates of the groups.

CameraData.csv (data on camera-trap detections of mandrills) was collected in a camera-trap survey in 2012–2014. We used circular kernel density estimation (with the R package "activity") to examine the temporal and seasonal variations in camera-trap detection rates of mandrill groups.

Supplementary_Rcode.R includes all the codes used in the analyses.