Inferring temporal variation in marine benthic communities from dead shell accumulations: A test using data from coastal biomonitoring

Relationships between the temporal dynamics of living bivalve communities and the formation of time-averaged death assemblages are explored in the Southern California Bight, where living assemblages in mid-shelf muddy sand and sandy mud habitats record substantial compositional changes during the last thirty years (sampled by the Southern California Coastal Water Research Project). Combining these data with dead sieve residues from a 2003 benthic survey, we can begin to assess how well death assemblages, assumed to be time-averaged over at least decades and subject to differential post-mortem destruction of species, capture (1) long-term community composition and (2) recent community states, and thus the utility of death assemblages for historical ecology. Living assemblages changed substantially between 1977 and 1998, owing largely to the decline in pollution-tolerant lucinoids, increasing abundance of Tellina spp., and increasing community evenness and richness.  We find that pairwise live-dead agreement is consistently lower than pairwise “live-live” agreement among successive live censuses (Bray-Curtis similarity and Spearman rho values both depressed by ~0.2), indicating that death assemblages are not simple summations of past community states. In sandy muds, differences between successive live censuses are relatively small; the death assemblage is positively correlated to all live censuses, but is most similar to the earliest, lucinoid-dominated community, implying “taphonomic inertia” to recent community change. In muddy sands, live-live agreement itself is more variable, and death assemblage patterns are complex. Simulations indicate that abundance of species that are highly abundant in death assemblages but rare in living communities cannot be explained either by demographic stochasticity or lifespan bias, but should be related either to temporal variations in population dynamics (most likely here) or to between-species durability differences. Thus findings of low live-dead agreement in benthic samples can indicate a substantial recent compositional shift.