The effect of survivorship bias on archaeological oyster valve size metrics: implications for fisheries baselines

Oysters (Ostreidae) are ecological and cultural keystone species that have been exploited by humans for millennia. Oyster size is a good proxy for population health, and researchers frequently use valve height and length measurements from archaeological and paleontological contexts that provide historical ecological baselines with potential to inform present-day management around the world. Oyster fragmentation in the archaeological record, however, is not commonly considered in studies of oyster size. This study presents the first systematic investigation of the effects of oyster fragmentation on eastern oyster ( Crassostrea virginica ) size metrics. Oyster assemblages including whole valves, valve fragments with hinges, and valve fragments lacking hinges were analyzed from stratified contexts at two Florida Gulf coast sites, Garden Patch (8DI4) and Calusa Island Midden (8LL45). Through comparisons of weight, height, length, and survival rates of whole valves, and the number and weight of fragmented valves, we show that survivorship bias—the selective preservation of whole valves of specific size ranges—explains significant variation in oyster valve size metrics. In some cases, survivorship bias is the primary driver of whole valve size differences between sampling units, indicating that whole valve samples are not representative of the pre-fragmented population. This conclusion has significant implications for current interpretations of archaeological oyster assemblage valve size as a proxy for past oyster reef health and effects of human predation and management. We suggest several methods for recognizing the confounding effects of survivorship bias in studies of archaeological oyster valve metrics and encourage continued critical evaluation of the archaeology of oyster and other shellfish assemblages and interrogation of their historical ecological implications in the present. • First study of fragmentation effects on oyster valve size representation. • Mean size of fragmented valves is estimated using total weights. • Identifies survivorship bias caused by unequal rates of fragmentation by valve size. • Whole valve samples are not representative of pre-fragmented populations.