Defining darkness: modelling decisions drive inference on moonlight effects in African lion nocturnal activity

Abstract Understanding how nocturnal species respond to variation in moonlight is central to chronobiology and behavioral ecology. Moonlight is the dominant light source at night in many ecosystems and can influence predator-prey interactions, navigation, and foraging behavior. However, research on its effects in free-ranging terrestrial mammals is hampered by inconsistent terminology and methodological approaches. Using GPS collar data from African lions ( Panthera leo ) in north-central Namibia, we examined how varying definitions and measurements of moonlight affect interpretations of nighttime activity. We reviewed six influential lion studies, extracted their moonlight-related predictors, and applied each approach to a standardized dataset to evaluate resulting variation in movement patterns (lion step length). We found that model outcomes varied substantially depending on how predictors were defined. Next, we developed and compared 15 candidate models to test how specific definitions – of night, lunar illumination, moonrise/set, and cloud cover – explain variation in lion step length across sex and seasonal strata. The best-performing models included three-way interactions between lunar illumination, moonrise/moonset, and cloud cover, but explained less than 1.2% of the variance in step length – urging caution in broad interpretations of the effect of moonlight on lion behaviour. Despite the weak direct effects, our results both underscore how divergent analytical choices can shape biological inference, and caution against simplistic explanations of lion nighttime activity. We recommend a standardized framework for modelling moonlight that distinguishes twilight from night, incorporates moon visibility, and treats measures of lunar illumination as continuous or cyclical. The recommended approach can enhance comparability and improve inference across studies of nocturnal species.