Clumped and oxygen isotope sclerochronology methods tested in the bivalve Lucina pensylvanica

Geochemical signatures preserved within the geologic record can be used to reconstruct past mean temperature and seasonality, but in order to accurately apply any geochemical proxy method in the past, a rigorous study of how the recorded proxy is related to temperature in the modern setting must be conducted. Here, we assess the ability of multiple isotope techniques to correctly record mean annual temperature and seasonality in the bivalve Lucina pensylvanica. We compare subannual-resolution δ18O-based, seasonally-targeted and continuous high-resolution (H.R.) clumped isotope (Δ47)-based thermometry methods, as well as multiple data treatment methods for each, to determine which approach best matches known modern temperatures (maximum and minimum absolute temperature and annual temperature range), with the goal of defining the ideal sampling scheme for use on fossil shells. In L. pensylvanica shells collected from 7 sites, we observe neither mean temperature nor seasonal biases. Mean annual temperature is best matched by averaging all seasonally-targeted Δ47-temperatures. Seasonality is best matched by averaging δ18Ocarb-based temperatures from all summers and all winters before taking the difference. Of two data treatment approaches applied to the continuous high-resolution Δ47-based temperatures, "data optimization" is apparently better at resolving smaller seasonal temperature differences. In contrast, "data smoothing" produces a temperature record unbiased by prior assignment of seasonal extremes and has the simultaneous ability to detect subannual variability in δ18Ow. However, accurate application of H.R. Δ47 methods must balance sampling resolution and growth rate. If sampling resolution is high enough relative to the growth rate (∼1 pt./month or better), we recommend continuous high-resolution Δ47-thermometry with data smoothing. If this resolution cannot be achieved due to slow growth rates or insufficient shell size, we recommend pairing subannual δ18Ocarb-based and seasonally-targeted Δ47-based temperature reconstruction to acquire seasonal range in temperature and absolute temperature extremes.