Correcting for durability bias in estimating the composition of death and fossil assemblages: implications for understanding past ecological changes

Dead-shell and fossil assemblages are tantalizing sources for information on past ecosystem-response to natural and anthropogenic perturbations. However, assessing the effects of interspecific differences in preservation on species abundances time-averaged dead-shell assemblages is hampered by the lack of long-term compositional data on present-day living assemblages. Here, we compare a unique, multidecadal-scale dataset of living assemblages (N~500,000) with surficial time-averaged death assemblages (N~40,000) from the southern California shelf to estimate the compositional bias that might arise from inherent differences in skeletal durability. We show that shell thickness is, among other four traits considered (shell size, mineralogy, organic content, and mode of life), the most important trait modifying molluscan species abundances. Using this as the null benchmark for live-dead discordance that is taphonomic in origin, the remainder of mismatch, such as the greater abundance of epifaunal suspension-feeders and siphonate deposit-feeders in death assemblages owes in fact to their ecological decline in recent centuries, even though their thicker shells and other attributes make them more preservable relative to thin-shelled remains of infaunal chemosymbiotic and detritus-feeders. Applying a correction factor informed by shell thickness sharpens the ability of live-dead mismatch to detect ecosystems modulated by human stressors, crucial to management and conservation decisions.