Figure 2: Evaluation of seasonality of carbonate precipitation
for biogenic, micrite, and biologically mediated lacustrine carbonates
using the clumped isotope calibration of Anderson et al. (2021). White
circles with black bars are Δ47-derived temperatures and
with uncertainties, respectively. Results are compared to projected lake
temperatures for different times of year, calculated using relationships
from (Hren & Sheldon, 2012) and Mering (2015) in conjunction with MAAT
for each location derived from the University of Delaware’s
high-resolution gridded air temperature dataset (Willmott & Matsuura,
2001) provided by NOAA. Orange circles reflect early spring (Northern
Hemisphere April-June (AMJ)) water temperatures, yellow circles indicate
summer (Northern Hemisphere June-August (JJA)), blue circles indicate
April-October water temperatures (A-O), and red circles indicate warmest
month water temperatures (WMT). This comparison shows that for biologic
carbonates, there is good agreement between Δ47-T and
what is known about typical growth seasons in the Anderson et al. (2021)
calibration in lower latitudes. However, the Anderson et al. (2021)
calibration results in an overestimation of growth temperature at higher
latitudes, with 37% of clumped isotope derived temperatures matching or
exceeding estimated lake warmest month temperature, as indicated by the
offset between the clumped isotope derived temperatures represented by a
white circle with black error bands and the colored data points in panel
A. For micrites, the Δ47 temperatures derived from the
Anderson et al. (2021) calibration are likely underestimated in roughly
half of the lakes examined here. Estimated temperatures from
biologically mediated carbonates using the Anderson et al. (2021)
calibration are reflecting warmest month temperatures in many cases, an
overestimation from the projections derived from Hren and Sheldon
(2012).
3.2 Δ47-Temperature Relationships
We utilize calibration data in Table 1 to derive regressions for our
entire dataset (a composite freshwater calibration), and
material-specific calibrations for biogenic carbonates (bivalves and
gastropods), biologically mediated carbonates (microbialites and tufas),
micrite, and travertines. Δ47 values for samples within
this study range from 0.409 to 0.682‰ with independently measured water
temperatures ranging from 5 to 95°C. Calibrations derived in this study
are presented in Table 2.
Performing an ordinary least squares regression through all freshwater
carbonate data in this study results in a steeper slope and shallower
intercept than Anderson et al. (2021) calibration. Although the 95%
confidence intervals on the estimated regression models overlap visually
(Fig. 3), an analysis of covariance (ANCOVA) analysis shows that the
slopes for the two calibrations are significantly different from each
other (pslope = 0.0334; Table 3). The Anderson et al.
(2021) calibration includes low temperature Antarctic microbialites that
are offset from other data, with half of the samples from a high pH
(10.3-10.7; Mackey et al., 2017) environment (Fig. 3 - dark red
symbols). Low temperature and high pH are environmental factors that
could give rise to potential kinetic isotope effects or DIC speciation
effects (Tang et al., 2014; Tripati et al., 2015). In fact, all of the
Antarctic samples are negatively offset from the rest of the data in
this study, and thus, we exclude these data from both the composite and
the biologically-mediated regressions we report in this study (Table 3).
Additionally, we compare our results from the composite calibration to
the synthesis of calibration data published in Petersen et al. (2019)
that was projected into a 90°C reference frame. Similarly to Anderson et
al. (2021), we find that the slope derived within this study yields a
statistically significant difference in slope to that of Petersen et al.
(2019) (pslope = 0.0036; Table 3). Despite improvements
in data handling and standardization procedures that have been shown to
resolve disparities in derived regression slopes in two large synthesis
studies (Anderson et al., 2021; Petersen et al., 2019), we do not
observe convergence between regression parameters derived from these
syntheses and our composite freshwater regression (Fig. 3). However, our
derived freshwater composite calibration parameters show agreement with
two previous calibrations using authigenic lacustrine carbonates (H. Li
et al., 2021) and travertines (Bernasconi et al., 2018) (Table 3).