CONFOUNDING EFFECTS ON DETERMINING TEMPERATURE FROM THE INTRASKELETAL CHEMICAL VARIATIONS OF MYTILUS CALIFORNIANUS
General linear models were used to assess the explanatory power of various environmental variables (temperature, δ18Oseawater, mussel size, intertidal position, chlorophyll a) for the observed intraskeletal δ18O and Mg/Ca variations. The predicted δ18O and intertidal position together explained 37% of the variation of the measured intraskeletal δ18O. Temperature and intertidal position explained 24% of the variation of the measured intraskeletal Mg/Ca with mussel size explaining an additional 5%. Tidal elevation differences include a δ18O enrichment relative to equilibrium at the higher intertidal height during warmer months due to (1) a possible growth slow/shut down at higher temperatures or (2) possible enrichment in δ18O above a certain temperature threshold. In addition, consistently higher Mg/Ca values at the higher tidal elevation can be explained by differences in temperature or potential feeding time between the two intertidal positions. Thus, M. californianus would not be a reliable tool to recreate past oceanographic conditions based on (1) the large amount of unaccounted variation from correlations with predicted variables (63% for δ18O and 72% for Mg/Ca), (2) confounding effects of intertidal position and specimen size, (3) the δ18O enrichment relative to equilibrium during summer at the higher intertidal elevation, and (4) the large number of replicate samples needed for an accurate temperature determination. Using additional data from previous studies, we demonstrate the precision of Mytilus as a paleotemperature recorder increases with percent submersion and, therefore, with lower tidal elevation.