The existence and persistence of rhythmicity in animal activity during phases of environmental change is of interest in ecology and chronobiology. A wide diversity of biological rhythms in response to exogenous conditions and internal stimuli have been uncovered, especially for polar vertebrates. However, empirical data supporting circadian organization of large ruminating herbivores remains inconclusive. Using year-round tracking data of the largest Arctic ruminant, the muskox (Ovibos moschatus), we modelled rhythmicity as a function of behaviour and environmental conditions. Behavioural states were classified based on patterns in hourly movements, and incorporated within a periodicity analyses framework. We found that ultradian rhythmicity was prevalent when muskoxen were foraging and resting in mid-winter (continuous darkness). However, the probability of rhythmicity declined with increasing photoperiod until largely disrupted in mid-summer (continuous light). Individuals that remained rhythmic during mid-summer foraged in areas with lower plant productivity (NDVI) than arrhythmic individuals. We conclude that muskoxen may use internal time keeping when forage resources are low, but that the importance of this mechanism weakens once environmental conditions allow energetic reserves to be replenished. We argue that alimentary function and metabolic requirements are critical determinants of biological rhythmicity in muskoxen, which likely applies to ruminating herbivores in general.

Movement data were collected using Global Positioning System (GPS) collars (Tellus Large; Followit Lindesberg AB, Sweden) on a total of 19 adult female muskoxen (n = 14 in 2013 and n= 5 in 2015). All GPS collars were programmed to record one position per hour and physically impossible movements were removed from the location data prior to analyses. 

Photoperiod for the study area was extracted from sunrise and sunset data downloaded from the U.S. Naval Observatory (http://aa.usno. navy.mil), elevation was extracted from the ASTER Global Digital Elevation Model (https://asterweb.jpl.nasa.gov/gdem.asp), and the location of dense vegetation areas was determined using Landsat 4-5TM satellite image. The dynamic variables NDVI and snow depth were extracted from the spatiotemporally explicit SnowModel and MicroMet modelling tools applied to the entire study area and period.

See ReadMe file for details.