Our initial article described the usefulness of Ultra-WideBands (UWB) set-up to study the individual activity and spatial organisation of 6 free-living indoor cats . In another but larger group of indoor cats (n = 14), the present study provides the first set-up combining UWB to record the amount of locomotor activity and passive RFID tags coupled with automated feeder to record eating behaviour. A chronobiological analysis of the continuously recorded data 24 h/24 over seven days allowed us to assess daily rhythmicity of locomotor activity and eating pattern in a colony of domestic cats living collectively and having ad libitum access to dry food and water.
The rest–activity rhythm is commonly studied using actigraphy, a non-invasive measure of circadian activity. The parameters that described rhythm characteristics included amplitude (A), period (P) and phase. Other analyses, such as interdaily stability (IS), intradaily variability (IV) and the least active 5 h (L5), had first been developed by Witting et al.  in order to study the effect of age and Alzheimer’s disease on rest–activity rhythm. These variables, called nonparametric as they are not associated with parameters of a known function, were later used by Van Someren et al. [35, 36] who described them as a more appropriate way to investigate and discriminate disturbed circadian rhythms, as well as sensitive to change even in small samples . The rhythms of the cat having previously been described as variable or irregular, these analyses seemed more fitted for our study. Besides, Piccione et al.  already used them to study the influence of different housing conditions on daily activity rhythm in cats.
Using precise and elaborate animal tracking methods, we mainly observed bimodality in the activity and feeding rhythms of these indoor cats with crepuscular peaks. Their bimodal patterns were not affected by sex or chronotypes. Hawking et al.  described many short bursts of activity distributed irregularly throughout the 24-h period, with no evidence of a daily cycle in a laboratory isolated cat. On the same year, Kavanau  agreed in classifying the activity of the cat as arrhythmic. Nevertheless, the very small sample size in both studies (n = 1) makes this conclusion doubtful. Recording the behaviour of more numerous individuals, we have demonstrated 24-h periodicity of locomotor and eating behaviour in every cat or almost every cat (11 out of 14), respectively.
The least active 5 h (L5), previously used only once on the activity rhythms in cats , have proved to be a pertinent rhythmic parameter (i.e. phase-marker) to compare in our study. They indeed corresponded to the main daily trough: the hours of the day when the cats were significantly moving and eating less coincided with the locomotor activity and eating L5, occurring either in the middle of the night, or in the middle of the day. The nocturnal decrease of activity in eight out of the 14 cats reminds the results of Hawking et al.  who noted the total duration of activity of a laboratory cat during light was about 1.4 times that of the activity during night during a 12 h light–dark cycle.
Besides, the rhythmic behaviours of eating and locomotor activity of the cats were notably impacted by twilight, as already observed in the literature [2, 5, 9, 11, 15, 17, 24]. The many authors who reported peaks of activity at dawn and dusk sometimes characterised these specific moments as key periods in biological and ecological processes as prey activity, food condition, body temperature, or colonic motility [4, 9, 11, 16, 26]. In our colony, when looking at the daily activity and eating graphs according to the hour of the day, consumption and activity peaks were indeed noted mainly preceding sunrise (between 04:00 h and 06:00 h) and sunset (between 16:00 h and 21:00 h), especially in the most active group (B). Positive modulators of these peaks also relied on anticipation of food renewal in the morning and end of human presence in the late afternoon. Yet, if the cats showed ultradian rhythms with period of 6 h or less rather than bimodal 24-h rhythms, peaks in the periodograms would then rise around 6 h or at lower periods. This was not the case, i.e. based on periodograms, the cats showed only significant peaks at 24 h and shorter ones at 12 h. Therefore, we conclude of two main daily peaks of locomotor activity and food consumption, corresponding mostly to anticipation of twilights (i.e. sunrise and sunset).
A great interindividual variability was observed in the eating and activity behaviour of this population, as regularly found in the literature (e.g. [21, 10, 26, 31]). The sex of the cats was a factor that significantly impacted their behaviour. The males, heavier than the females, tended to be more active and were more rhythmic in their locomotor behaviour as shown by a trend for larger daily amplitude and lower interdaily stability. They ate significantly more than the females and none of them were arrhythmic in their eating behaviour, contrary to three females out of seven. When looking at the mean hourly covered distance and consumption, we noticed that the sex differences only happened at the peak hours, while both males and females diminished their activity and consumption simultaneously in the middle of the day and middle of the night.
Circadian rhythmicity can only be measured in constant conditions of light or darkness. Such study conditions enable the detection of endogenous behaviour but can erase external factors determining the natural patterns in animals by not matching natural environmental conditions. Here, we measured daily rhythmicity in indoor rooms where ambient humidity and temperature were kept constant, but natural daylight was received through several bay windows. Although this environment prevented us to generalise our results to cats living in outdoor conditions, it highlighted the impact of daylight fluctuations, separated from other environmental daily variations such as in ambient temperature and humidity, on the locomotor and feeding rhythms of the cats we studied. Moreover, the similarities we observed between the behaviour of the cats of these studies compared to cats living in wilder conditions, such as bimodality and crepuscularity in their rhythms [4, 7] demonstrate some factors determining their natural behaviour were still present in the indoor environment they lived in for the present study.
Based on actogram characteristics, L5 values, and nocturnal and diurnal activity/consumption rates, we attempted to further investigate differences among individuals by categorising them according to typical dominant chronotypes: diurnal versus nocturnal. The main difference between diurnal categorised individuals and nocturnal categorised individuals resides in the more pronounced peaks of activity and consumption at twilights in the nocturnal categorised ones, while the activity and consumption troughs are similar between the two types of individuals. Therefore, the chronotype categorisation, which has been commonly used in other species, is difficult to highlight in the cats, which echoes the variable and contradictory categorisations in the literature (e.g. [12, 19, 37]). Refinetti et al.  noticed the chronotype spread (i.e. a measure of the variability of chronotypes among individuals) was the greatest in cats (23 h) compared to 15 other species, concluding the diurnal–nocturnal dichotomy should not apply to cats, as some individuals are mostly active at day while others are mostly active at night. Furthermore, for Refinetti , an absence of daily rhythm in the body temperature of the cat as found in the study of Hawking et al.  indicates how the cat distinguishes itself from pure nocturnal species or diurnal species for which the body temperature reaches its acrophase, respectively, during the night or during the day.
No general pattern emerged according to the chronotype categorisation. Nevertheless, regular peaks in the feeding rhythm, and mostly in the activity rhythm of the cats, have decidedly been demonstrated in every cat, along a 24-hours periodicity. More precisely, most cats showed two main peaks and troughs in the day, different from unimodal circadian rhythms where only one main peak and one main trough are usually demonstrated. Bimodality, more than chronotypes, seems therefore the best way to categorise the activity and feeding rhythms of the cats. Accordingly, Randall et al.  suggested that “two peaks may be the one factor that is common to the idiosyncratic patterns of entrainment in this species” and Refinetti et al.  detected bimodality in the activity rhythm of cats while chronotypes varied greatly among individuals.
It is worth noting that some individuals—the nocturnal categorised ones—ate practically not at all in the middle of the day (between 11:20 and 15:00 h). This highly pronounced trough, studied on a 7 days collection period, should constitute a pertinent feature for the eating behaviour of this species.
Systematically, the locomotor behaviour of the cats was more rhythmic than their eating behaviour: the activity rhythm of the cats was proved to have higher rhythm amplitude, interdaily stability, less intradaily variability, and every cat showed a 24-h periodicity in their activity rhythm, while three cats were arrhythmic in their eating behaviour. This difference in rhythmicity may explain the less pronounced differences between categories in the feeding rhythm of the cats, than in the locomotor activity rhythm. Besides, this observation may be due to the opportunistic nature of this solitary hunter which, in the wild, feeds on several small prey per day with various rhythms (notably, diurnal birds and nocturnal rodents) and may therefore display flexibility in its eating patterns to adapt to the daily rhythms of its prey, as suggested by Konecny .
One difficulty in the analyses of this study relied in group differences. The mean daily covered distance differed significantly between them, group B being more active. This may be explained by the fact that individuals of group B had been living together for 5 years, whereas two cats of group A were new to the group. Difference in activity rate might be due to avoidance, a common aspect in a solitary species in spite of large inter individual variation in tolerance: less tolerant individuals spending more time in avoiding other conspecifics.
Along with this difference in covered distance, a group effect affected sex and chronotype ratios: most cats of group B were males, most individuals of group A were females and males and females were mainly classified as nocturnal and diurnal, respectively. Therefore, we could not cross the sex and chronotype effects since it was not equally distributed, and could not distinguish the sex effect, the group effect and the chronotype effect from each other. Mainly, the males, nocturnal and from group B were more rhythmic and active than the females, diurnal and from group A. Nevertheless, this group difference is interesting as it underlines the plastic behavioural characteristic of this species, able to accommodate to group behaviours . Furthermore, it would be interesting to establish whether intra-individual plasticity is stable or more variable with longer durations of data collection. It is noteworthy that cats of a same group showed very similar rhythms, while cats between the groups were more different. This social aspect of the species living collectively may confer an evolutionary advantage as it allows the individuals to adapt to various living conditions.
Variations in activity and food intake may be sensitive to exogenous factors, such as temperature or photoperiod, and to human intervention. In our conditions, we could not standardise human activity due to time constraints imposed to technicians. This might have differentially affected the activity of the cats. A greater sample size and a standardisation of human activity might help in search of a daily rhythmicity in the activity of the domestic cat.
Moreover, it would be of certain interest to compare the rhythmic behaviours of indoor cats, such as those of this study, with the behaviours of individuals having access to outdoor environment. Differences in housing conditions were proven to result in distinct daily activity rhythms: a group of pet cats having low access to an outdoor garden lived in stronger symbiosis with their owners and was most active during the photophase (daytime) compared to individuals having outdoor access more often and which exhibited more robust daily rhythmicity and the highest level of activity during the scotophase (nighttime, ). The results of the present study should, however, be compared with observations from similar activity tracking methodology, i.e. data from daily covered distance, contrary to studies using accelerometers technologies (such as ), for example.