Publikationsrepositorium - Helmholtz-Zentrum Dresden-Rossendorf

Animal encounters are key components of population dynamics, community dynamics, and
evolutionary processes. Consequently, measuring encounter rates (i.e. encounters per time) can be
insightful. Encounter rates can be measured from animal tracking data, using metrics that can be split
into two groups. The first group consists of trajectory-based metrics, i.e. measures based on serial
records of animal locations. This first group includes PROX, the number of observed per number of
samples. The second group, in contrast, consists of metrics based on home range overlap, including
the Bhattacharyya coefficient (BC). In this study, we argue both types of metrics are limited.
Trajectory-based metrics are direct measures of encounter rates but have statistical estimation
issues due to their dependency on the frequency of location sampling. Meanwhile, home-rangebased metrics are statistically sound but are not proportional to encounter rates. To overcome both
challenges, we proposed a new metric, Relative Encounter Rate (RER). RER increases linearly with the
number of encounters and does not depend on the frequency of sampling (i.e. it is scale-invariant). In
an individual-based simulation, we measured how RER, BC, and PROX relative error under different
sample sizes and sampling frequencies. Further, we compared these metrics in three empirical case
studies. We tested Jaguars for polygyny, deforestation effects on tapir connectivity, and an extension
of the dearest enemy hypothesis with brown bears. We also compared partner hierarchy according
to BC and RER in Jaguar mating clusters. In the simulation study, we found PROX overestimates the
encounter rate when data has a low sampling frequency. The simulation also indicates BC
overestimated encounters. Furthermore, PROX led to false positives in the Tapir and Bear case
studies. In addition, PROX was incapable of detecting many individual relationships in the jaguar
polygyny study. RER does not depend on sampling frequency (contrary to PROX) or sample size
(contrary to BC). We discuss further hypotheses to test with RER and argue RER can enable ecologists
to analyze encounters with a level of detail adequate to their importance, leading to a better
understanding of how individual behaviors influence population and community dynamics.