The intensity of biogenic sediment mixing is often expressed as a "biodiffusion coefficient" ( Db), quantified by fitting a diffusive model of bioturbation to vertical profiles of particle-bound radioisotopes. The biodiffusion coefficient often exhibits a dependence on tracer half-life: short-lived radioisotopes (e.g. 234Th) tend to yield notably larger Db values than longer-lived radioisotopes (e.g. 210Pb). It has been hypothesized that this is a result of differential mixing of tracers by particle-selective benthos. This study employs a lattice-automaton model of bioturbation to explore how steady-state tracers with different half-lives are mixed in typical marine settings. Every particle in the model is tagged with the same array of radioisotopes, so that all tracers experienced exactly the same degree of mixing. Two different estimates of the mixing intensity are calculated: a tracer-derived Db, obtained in the standard way by fitting the biodiffusion model to resulting tracer profiles, and a particle-tracking Db, derived from the statistics of particle movements. The latter provides a tracer-independent measure of mixing for use as a reference. Our simulations demonstrate that an apparent Db tracer-dependence results from violating the underlying assumptions of the biodiffusion model. Breakdown of the model is rarely apparent from tracer profiles, emphasizing the need to evaluate the model's criteria from biological and ecological parameters, rather than relying on obvious indications of model breakdown, e.g., subsurface maxima. Simulations of various marine environments (coastal, slope, abyssal) suggest that the time scales of short-lived radioisotopes, such as 234Th and 7Be, are insufficient for the tracers to be used with the biodiffusion model. 210Pb appears an appropriate tracer for abyssal sediments, while 210Pb and 228Th are suitable for slope and coastal sediments.