Reconstruction of the temperature history of individual fish would be of considerable value to those studying growth trajectories, spawning sites and migration pathways, among other things. In principle, the chemical composition of the daily growth increments formed in the otolith records the temperature history of the fish on a daily basis. Based largely on previous studies of coral, two approaches have been employed for otolith-based temperature reconstruction: one based on the ratio of the elements strontium and calcium, and the other based on oxygen isotope ratios. Early applications of Sr:Ca ratios were promising, and appeared to indicate a reasonable relationship with temperature. However, subsequent research has indicated that there is no generalized relationship between Sr:Ca and temperature, with the possible exception of cold-water larvae. Therefore, attention has shifted to analyses of oxygen isotope ratios.

The basis for temperature reconstructions is that oxygen incorporated into carbonates (such as the otolith) includes both major isotopes (¹⁸O and ¹⁶O). Normally, one would expect them to be deposited in equilibrium with their concentration in the water. However, there is a physical fractionation due to temperature at the time of deposition, such that the proportion becomes increasingly depleted as temperature increases. As a result, if the ¹⁸O : ¹⁶O in the ambient water is known, and if the ¹⁸O : ¹⁶O in the otolith can be measured, temperature can be calculated. With the recent development of otolith micromilling devices, small regions of the otolith (~ 50 µm in width) can be sampled and assayed, thus providing temporal resolution on the order of months.

The major constraint to the application of oxygen isotope assays is the requirement for knowledge of the ambient ¹⁸O : ¹⁶O in the water during the period of interest. Where this is known, otolith assays should provide an accurate and unambiguous temperature history of the fish. If the isotope ratio of the water is unknown, salinity can often be used as a proxy for the ratio. However, the temperature history may then become a relative one, rather than absolute. In those instances where fish migrate through various water masses characterized by large, unmeasured differences in oxygen isotope composition, reconstruction of the temperature history is likely to be flawed. Nevertheless, the microsampled isotope ratios will still record the migration through the various water masses on an age-structured basis.

An experiment demonstrating the value of oxygen isotope ratios as a proxy for temperature is presented in Thorrold et al. (1997b). A complete review of the field, including the presentation of a new isotope-temperature fractionation equation, is presented in Campana (1999).  Jones and Campana (2009) use oxygen isotope ratios to reconstruct the temperature history of a collapsed cod population, demonstrating that cod thermoregulated, but did not leave the area, when engulfed in a cold water mass.