A collision between wildlife and vehicle (WVC) is a phenomenon comparable to others with points patterns, like traffic accidents, crimes, cases of epidemic diseases or extreme climate events. Knowing the aggregation patterns of these events is one of the main challenges of those areas researchers (Miller & Han, 2009, Liu et. al. 2015). Aggregation can be based on a proximity of events in space, time or, simultaneously, in space and time (Miller & Han, 2009). Determine the sites with non-random spatial aggregation of WVC (hotspot) is essential for the design of mitigation measures (e.g. Beaudry et. al., 2008, Malo et. al., 2004). Some studies indicate that hotspots are spatially stable in generation (Clevenger et. al., 2003, Carvalho & Mira, 2010) while others state that their persistence may vary over time. Permanence is due to the biological characteristics of the studied group (Langen et. al. 2006, Costa et. al., 2015) or insufficient sampling frequency (Santos et. al., 2015). The realization of hotspots temporal variations, as well as the explanations for the occurrence of those variations, do not elucidate in which sites measures to prevent WVC should be installed. Identify places where the high probability of collision persists in time increases the feasibility of mitigation measures installation and their efficiency on wildlife conservation. It is important though to know the spatial and temporal patterns of WVC simultaneously. Since the middle of last century there are studies on spatiotemporal clustering patterns in punctual events, particularly in epidemiology and criminology, and more recently in traffic engineering (Eckley & Curtin, 2013). Despite this knowledge availability, space-time interaction has not been emphasized in Road Ecology literature. We reviewed 56 articles published between 2001 and 2015 to assess the importance given in Road Ecology to studies approaching the space-time interaction in the analysis of aggregate WVC. Thirty-two of these articles evaluated only spatial patterns, 9 tested only temporal patterns, and 15 considered the spatial and temporal patterns. However, regarding this third group, only Mountrakis & Gunson (2009) attempted to estimate the association linking the temporal and spatial scales simultaneously. The other authors considered both scales, but in different analyzes. Those authors used an adapted Ripley’s K statistic (Diggle et al., 1995) to assess which spatiotemporal scales indicate aggregation of Moose-Vehicle collision on roads of Vermont, USA. Albeit they have not shown the places where such points of aggregation occurred, they demonstrated that the use of space-time interaction generates more robust results because it can recognize aggregation or dispersion caused by sporadic events. Notwithstanding indicating the relevance of this approach, neither the articles that cited this study applied a spatiotemporal dependence analysis of WVC aggregation. Road ecology researchers should pay more attention to the spatiotemporal relationship of the roadkills’ aggregation, to diminishing the effect of spatially or temporally stochastic events in the hotspots' identification. The analysis of spatiotemporal interaction on punctual events can support the identification of sampling sufficiency to ensure the spatial persistence of hotspots in the appropriate spatial scales regarding the subjects of conservation.
Ripley´s K; mitigation; hotspot; sampling sufficiency; scales