Movement is a pervasive ecological process that influences most facets of animal life. Studies investigating the environmental drivers of movement have increased in abundance in recent years (Long and Nelson 2013), with numerous studies applying a statistical approach to model the relationship between the moving object and the movement space. However, the definition of the moving object and movement space are subject to a number of conceptual and methodological challenges. Laube (2017) recently described seven semantic levels of quantifying movement that ranged from a local time-stamped location to a global aggregation demonstrating the probability of space use. Subsequently, how movement is defined can have substantial implications for inferences made from analysis. For example, Holloway and Miller (2018) found that the area of landscape that was accessible to simulations of 25 brown hyenas over a year period was over 60,000km2 larger when movement was measured as ‘moves’ compared to ‘fixes’. Moreover, the definition of the moving object cannot exist independently of the movement space. The environment is inextricably linked with movement (Nathan et al. 2008), and with spatial statistics increasingly using environmental covariates to infer movement-environment relationships, we need to revisit how define the environment relative to the moving object.
In this study, I use step-selection functions (SSF - a spatial statistic that analyses movement patterns in relation to the underlying environment compared to alternative movements that the object could have taken - Fortin et al. 2005) to investigate how the choice of representation of movement in relation to the movement space affects animal-environment inferences. By systematically altering the conceptualisation of movement for a number of species, I found that these decisions strongly influence the results of the SSF and subsequent inferences about animal movement and environmental interactions. Significant differences were consistently found for relationships investigating a directional persistence (e.g. towards or away from a landscape feature), with movements defined as ‘moves’ or ‘vectors’ often finding a significant relationship, with other conceptualisations of movement finding no significant relationships. Differences were found among almost every definition of the moving object and movement space used in the analysis. The results of this research should foster discussion across a multitude of disciplines including movement ecology, ecological modelling, GIScience, and spatial statistics.