Assessing attraction or avoidance between rabbits: Comparison of distance-based methods to analyse spatial distribution

Studies on social spacing usually analyse spatial distribution using a single distribution index, and compare this with expected values assuming random spacing. However, different distribution indices give information on different aspects of spatial distribution. Furthermore, when animals use some parts of the study area more often because of environmental factors, assumptions of random spacing do not apply. As a contribution to the development of spatial distribution analysis methodology, we compared three distribution indices: nearest-neighbour distances (NND), coefficient of variation of interindividual distances and coefficient of variation of Dirichlet polygon areas. Distribution was studied at different animal densities, in the presence or absence of a wooden structure. Rabbits, Oryctolagus cuniculus, were chosen as the model species because of their apparent plasticity of spatial behaviour. The observed distribution was compared with a random distribution and with a resource-corrected random distribution that accounted for environmental influences on spacing. Comparisons with these two distributions led to different conclusions. This emphasizes the limited usefulness of comparisons with the random distribution, especially in more heterogeneous environments. However, divergence from random spacing could potentially be used to assess the relative importance of resources. Especially for 6-week-old rabbits, the density at which no evidence of avoidance was found differed between the distribution indices. This underlines the need to use multiple indices when studying social spacing. Nine-week-old rabbits generally avoided each other's proximity, with NND indicating proximity avoidance even in the largest pens. Contrary to expectations, the wooden structure seemed to decrease social attraction between rabbits.