Stable isotope-based proxy methods enhance our ability to interpret geodynamical histories for tectonic provinces via paleoelevational reconstructions. These methods require that the unmodified isotopic composition of meteoric water is recorded by authigenic minerals, a critical assumption that has not been tested across wide-ranging environmental and topographical contexts. Here, we show that Quaternary lake carbonateδ18O values are not strongly, nor significantly, correlated with regional elevation due to the isotopic modification of in-flow waters following entry into the lake environment. These modifications are largely caused by surface water evaporation, and can result in >3 km errors in paleoelevation estimates if not accounted for. However, our analysis suggests that positive shifts in surface waterδ18O are accompanied by similar magnitude shifts in δ13C-DIC. This positive co-variation inδ18O and δ13C may be used to detrend lake carbonate compositions for the effects of surface water evaporation using a parameter we define here as the ‘13C-excess’. When Tibetan lakes are excluded from the data set,13C-excess values are significantly correlated with mean up-slope hypsometric altitude with an error of ±500 m. Application of the13C-excess approach to Cenozoic lake carbonate records from the western U.S. Cordillera both challenges and reinforces previous paleoelevational interpretations based onδ18O alone.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 This shortcoming of stable isotope-based paleotopographic methods is particularly problematic for authigenic minerals that formed in lakes due to the isotopic effects of surface water evaporation. Yet, lake carbonates are some of the most widely applied mineralogical proxies for past elevations [Norris et al., 1996; Garzione et al., 2000; Horton et al., 2004; Horton and Chamberlain, 2006]. The widespread publication of Quaternary lake carbonate isotopic compositions presents a unique opportunity to test the assumption that lake carbonates accurately reflect meteoric-derived water compositions.
 We compiled 13,066 authigenic carbonate δ13C and δ18O values from 27 Quaternary lake systems spanning a ∼5000 m elevation range, a >90° latitudinal range, and a variety of physiographic settings (see Table 1 and references therein). All samples are less than 155,000 years old, and most are Holocene in age (Table 1). We have excluded all samples explicitly described as Chara, charaphyte encrustations, or any variety of biogenic shell carbonate. Modern mean up-slope hypsometric altitudes were calculated for 250 km radius circles centered on each of the lakes studied using 3 arc-second elevation (±10 m accuracy) raster data using GIS software. All stable isotopic data was normalized to calcite in a coastal 45° latitude environment of formation at 25°C using published equilibrium fractionation equations [Kim and O'Neil, 1997; Mook et al., 1974; Bottinga, 1968] and modern lake surface water temperatures, a second-order polynomialδ18O versus latitude fit for a global meteoric water data set [Bowen and Wilkinson, 2002], and a meteoric water continentality δ18O lapse rate of 0.002‰ km−1 [Criss, 1999]. These normalizations serve as a correction for the first-order isotopic effects of latitude, continentality (measured as the shortest distance to the coast) and temperature, thus isolating the effects of elevation and evaporation on authigenic lake carbonate isotopic compositions. For comparison, we also calculatedδ18O values for ‘in-flow carbonates’ using published isotopic compositions for surface waters flowing into the lake systems studied. In-flow carbonateδ13C values were also calculated where sufficient data was available from the primary literature. All calculated in-flow carbonate values were normalized in the same way as measured lake carbonate values.
Table 1. Quaternary Lake and Sample Information, Isotopic Corrections for Latitude, Continentality, and Temperature Normalization, Lake Inflow Water Isotopic Compositions, and Primary References
3. Lake Carbonate Oxygen Isotopes, Elevation, and Evaporation
 The compiled data set in Figure 1 shows that mean normalized Quaternary lake calcite δ18O values are weakly and not significantly correlated with regional mean up-slope hypsometric altitudes (r2 = 0.20; n = 27; SEaltitude = 1,417 m; α= 0.01; p > 0.01; t-test). This result is not surprising as lake waterδ18O values are also not strongly correlated with elevation [Henderson and Shuman, 2009]. Based on these results, lake carbonate oxygen isotope compositions appear to be poor proxies for modern regional elevations. However, when Tibetan lakes (n = 7) are removed from the data set, a stronger and highly significant (r2 = 0.51; n = 20; SEaltitude = 908 m; α= 0.01; p < 0.01; t-test;Figure 1) correlation between lake carbonate δ18O values and regional elevations is present. These conflicting results raise the question: Why are Tibetan lake carbonates such poor proxies of regional elevation?
 At the regional scale, modern Tibetan systems fail to satisfy the basic assumptions of stable isotopic paleoelevation proxy research. Average annual meteoric water δ18O values across the Tibetan Plateau range between −5.5‰ in the northwest and −15.3‰ in the southeast, despite its consistently high elevation and relatively low relief [Araguás-Araguás et al., 1998]. Seasonal shifts in meteoric water isotopic compositions in many locations are even larger [Wei and Gasse, 1999], and Quaternary lake carbonates from the plateau range between ca. −14‰ and +6‰ δ18O (V-PDB) [Yu and Ricketts, 2009]. Within the Tibetan Plateau itself there is no correlation between altitude and meteoric water δ18O (r2 = 0.0093), despite a moderately strong correlation (r2= 0.63, second-order polynomial) south of the Himalayan crest [Quade et al., 2011]. Although a detailed analysis of the anomalous isotopic behavior that characterizes the Tibetan hydrosphere is beyond the scope of this paper, a brief discussion of the likely causes of these results is worthwhile.
 The highly variable isotopic compositions of modern precipitation, surface waters, and lake carbonates across the Tibetan Plateau result from the complex atmospheric and hydrologic processes and conditions affecting the region. These processes include: 1) competing influences of the isotopic amount and temperature effects on meteoric water compositions; 2) variability in the extent of Rayleigh distillation due to changes in meteoric inputs derived from converging air parcels following disparate trajectories; 3) highly variable catchment-scale hydrology and associated mixing effects; 4) spatial differences in the amount of seasonal evaporation due to the combined effects of wintertime ice-over duration and summertime cloud cover; 5) the extent of moisture recycling across the region [Araguás-Araguás et al., 1998; Bershaw et al., 2012; Kurita and Yamada, 2008; Wei and Gasse, 1999; Yu and Ricketts, 2009]. It is our opinion that lake carbonate isotopic records from this region must be interpreted with extreme caution until these factors, and their relative importance across different spatial and temporal scales, are more fully understood. Thus, due to the complex nature of the Tibetan hydrosphere, the absence of a significant correlation between altitude and the isotopic composition of meteoric waters across the region, and the lack of a general consensus as to the cause of these patterns, we have excluded Tibetan lake systems from our analysis.
 Yet, even when Tibetan lakes are excluded from the global database we compiled, carbonate oxygen isotope compositions still do not directly reflect regional meteoric-derived lake in-flow water compositions. This is a crucial observation relevant to stable isotopic paleoelevation research in any context.
 To demonstrate this point we present calcite δ18O values, calculated from published lake in-flow waterδ18O values and normalized as described above, together with normalized lake calcite δ18O values (Figure 1). Calculated in-flow calciteδ18O values are, on average, 6.8 ± 3.5‰ (±1σ) more negative than measured lake calcite values. The only realistic mechanism for such large isotopic modifications is evaporation. Kinetic fractionation during evaporation enriches lake waters in 18O (and deuterium) relative to in-flow waters, and the magnitude of these fractionations can exceed 15‰ (δ18O) [Henderson and Shuman, 2009]. As stable isotopic proxies of paleoelevation are most useful in orographic rain shadows [Chamberlain and Poage, 2000; Poage and Chamberlain, 2001], meaningful interpretations of regional paleoelevations using altitudinal isotopic lapse rates determined for meteoric-derived waters must take evaporative effects into account [Henderson and Shuman, 2009].
4. The 13C-Excess Parameter as an Elevation Proxy
 In addition to modifying lake water δ18O, evaporation is also known to abiotically enrich dissolved inorganic carbon (DIC) in 13C [Stiller et al., 1985]. A recent experimental study further demonstrates that positive linear co-variant trends betweenδ18O and δ13C are produced during evaporation-induced carbonate precipitation [Ufnar et al., 2008], and isotopic co-variation is a commonly applied proxy for hydrologic balance [Talbot, 1990]. Co-variant trends can also result from isotopic disequilibrium in authigenic carbonates formed as the result of rapid CO2 degassing [Hendy, 1971], a possible cause of isotopic co-variation in lakes with elevated CO2 concentrations relative to local atmosphere, or in systems where carbonate precipitation and associated CO2degassing is fast. Increases in net primary productivity during summer months can also drive isotopic co-variation in carbonate lakes due to the combined effects of discrimination against13C during photosynthesis and increased evaporation during drier periods [Talbot, 1990]. Most, if not all, of these processes are likely to occur in lake systems downwind of topographic barriers; yet, it is not known how, if at all, isotopic co-variant trends are related to regional altitudes.
 When classified by regional mean up-slope hypsometric altitude, the compiled Quaternary lake carbonate isotopic compositions plot along positive co-variant trends originating from more negativeδ18O values in higher altitude regions and more positive δ18O values in lower altitude regions (Figure 2). This patterned distribution results from the combined effects of altitude and evaporation on the isotopic compositions of meteoric-derived in-flow waters. Based on this empirical pattern, we hypothesize that y-intercept values ofδ13C versus δ18O positive linear co-variant trends are not significantly correlated (i.e., p ≥ 0.01) with regional mean hypsometric up-slope elevations.
 To test this hypothesis, we determined line equations relating average lake carbonate δ18O and δ13C values to calculated in-flow calcite values. In-flow calcite values (Figure 2) were determined using published lake in-flow waterδ18O and δ13C-DIC (as HCO3−) values and normalized in the same way as lake carbonates (Table 1). Using the average of the resultant line-equation slopes (m) of 1.0 ± 0.3 (±1σ), y-intercept values were calculated according toequation (1) and are here defined as 13C-excess values:
where, δ18Oavg is the average of the latitude, distance, and temperature normalized lake carbonate δ18O values, δ13Cavg is the average of the temperature normalized lake carbonate δ13C values, and 13Cex is the 13C-excess.
 Our analysis demonstrates that 13C-excess values are significantly correlated with regional mean up-slope hypsometric altitudes (r2 = 0.84; n = 18; SEaltitude = 504 m; α= 0.001; p > 0.001; t-test;Figure 3). When compared to the correlations presented in Figure 1, these results indicate that the 13C-excess approach is a more statistically robust and more precise paleoelevation proxy than lake carbonateδ18O alone, when Tibetan lakes are excluded. However, it is crucial to denote the assumptions underpinning the 13C-excess approach: 1) latitude, distance from coast, regional atmospheric circulation patterns, and temperature of formation must all be reasonably well known; 2) isotopic equilibrium conditions were achieved at the time of carbonate formation; 3) lake waterδ18O and DIC-δ13C are proportional to in-flow waterδ18O and DIC-δ13C according to a co-variant trend, wherem = 1. The first two assumptions apply to all authigenic carbonate stable isotopic proxies of paleoelevation [Chamberlain et al., 2012], while the third assumption provides a quantitative basis for detrending the effects of evaporation on lake carbonate isotopic compositions.
 We believe the assumption that lake water δ18O and DIC-δ13C are proportional to in-flow waterδ18O and DIC-δ13C according to a co-variant trend, wherem = 1, is justified for the following reasons: 1) the slope, m= 1, represents the average co-variation slope linking calculated in-flow derived carbonate compositions to lake carbonate compositions using data available in the primary literature (n = 7 lake systems); 2) the value is supported by experimental results. We expand on these points below.
 Paramount to this discussion is the recognition that for the purposes of stable isotope-based paleoelevation research, it is the isotopic composition of meteoric derived in-flow waters that is desired and not the isotopic composition of the lake waters themselves. Lake water, and thus lake carbonate,δ18O or δ13C values are known to be affected by: 1) changes in net primary production; 2) temperature and hydrological balance; 3) the extent to which carbonate formations is biologically mediated; 4) carbonate crystallization and carbon dioxide evasion rates; 5) shifts in lake-water source areas. These processes will also influence the direction (i.e., sign), statistical significance, and slope of isotopic linear co-variation among lake carbonate samples. Althoughδ18O and δ13C isotopic linear co-variation is strong (r2> 0.5) in several of the data sets we compiled, the slopes of these co-variant trends are not significant (t-test, p > 0.05). In other words, simple linear regression is a poor model for isotopic co-variation among lake carbonates. More important to stable isotope paleoelevation research, however, is the demonstration that the cumulative effects of the in-lake processes listed above are relatively minor when compared to the magnitude of the isotopic modification of in-flow waters following entry into the lake environment (Figure 4).
 For example, although the Quaternary carbonate isotopic record for Mono Lake (California, U.S.A.) does not exhibit an overall positive co-variant trend betweenδ18O and δ13C (r2 = 0.001; Figure 5), lake carbonate compositions are all enriched in 18O and 13C relative to carbonates formed from unmodified in-flow waters (Figure 5). The co-variation slope linking in-flow to the average lake carbonate composition is 1.2, equating to a ∼200 m higher regional hypsometric altitude than would be calculated using a co-variation slope of 1.0 (linear fit,Figure 4). Using co-variation slopes of 2.1 and 0.7, the slopes of the lines passing through the calculated Mono Lake in-flow carbonate value and tangent to the Mono Lake carbonate 95% confidence ellipse, the calculated altitudes differ from them= 1 value by +1200 m and −400 m, respectively. In other words, despite the complexity of in-lake processes and conditions, evidenced by the wide range in lake carbonate isotopic compositions and the lack of any meaningful isotopic co-variation in the lake itself, the large differences between in-flow and lake water chemistries allows for reasonable sub-kilometer scale approximations of the regional hypsometric altitude using the13C-excess approach. As the Mono Lake example demonstrates, some precision (i.e., kilometer-scale errors) is sacrificed when extreme lake carbonateδ-values, not representative of the true mean composition, are used to calculate a range in co-variation slopem values. The much more important point here is that, in contrast to the 13C-excess approach, the traditionalδ18O altitude proxy would underestimate the regional hypsometric altitude of the Mono Lake region by >3 km due to the false assumption that the authigenic lake minerals were functioning as a direct proxies of regional meteoric water isotopic compositions.
 The above Mono Lake example is not exceptional. Comparing the standard deviations of lake carbonate δ-values to the differences between average lake carbonate isotopic compositions and calculated in-flow carbonate compositions reveals that the isotopic modification of in-flow waters is up to a factor of ten larger than the variation caused by in-lake conditions (Figure 4). In all of the systems we studied, lake carbonates are enriched in 13C and 18O relative to in-flow water derived carbonate. As the isotopic effects of in-lake processes are relatively small compared to the large shifts in in-flow water compositions, and the isotopic modification of in-flow waters is always toward more positiveδ-values, the resultant positive co-variant trends between in-flow and lake compositions provide an empirical basis for the13C-excess approach.
 Unfortunately, few investigations have analyzed lake in-flow waters for temperature,δ18O and δ13C-DIC despite the relative abundance of published lake carbonate studies. For the seven we did find in the primary literature, the average linear co-variation slope, linking calculatedδ18O and δ13C values in equilibrium with in-flow waters to average lake carbonateδ18O and δ13C values, is 1.0 ± 0.3 (±1σ, Table 2). The available data suggest a systematic quantitative pattern between in-flow and lake water isotopic compositions is present. This observation raises two important over-arching questions: 1) What processes are causing in-flow and lake water compositions to co-vary? 2) How robust is the observed in-flow/lake co-variation slope ofm = 1? Although, complete answers to these questions are not immediately apparent in the literature, several points can be made in response to both questions.
Table 2. Normalized Quaternary Lake Isotopic Compositions and Covariation Parametersa
Number of Carbonate Analyses
Average δ18O (‰)
Standard Deviation δ18O (‰)
Average δ13C (‰)
Standard Deviation δ13C (‰)
δ18O vs. δ13C Lake Carbonate Slope
Lake Carbonate r2δ18O vs. δ13C)
δ13C Standard Error
δ18Olake − δ18Oin-flow
δ13Clake − δ13Cin-flow
δ18O vs. δ13C Lake: In-Flow Slope
All δ-values are relative to the V-PDB scale. Slope and coefficient of determination values are based on linear-regressions.
Lake located on, or adjacent to, the Tibetan Plateau.
Lake not included in analysis due to source area or geothermal effects (see text).
 Lakes are known to exhibit highly variable pCO2, and total DIC concentrations, with statistically significant (p < 0.01) positive co-variations betweenδ13C-DIC values and lake water alkalinity, pH, and surface area [Bade et al., 2004]. The waters that flow in to lakes tend to be depleted in 13C due to the presence of organic-matter-derived dissolved carbon. As these waters equilibrate with atmospheric carbon dioxide, theδ13C-DIC values will increase due to the combined effects of kinetic fractionation associated with CO2evasion and slower equilibrium fractionation reactions, much like the carbon isotope systematics of cave drip-waters and resultant speleothems [Hendy, 1971]. The kinetic and equilibrium fractionations driving enrichment of DIC in 13C will also affect the δ18O-DIC values (and thus theδ18O-H2O through equilibrium exchange); however, the small size of the δ18O-DIC reservoir in comparison to the abundant oxygen present in the water itself allows these oxygen isotope effects to be ignored. Rather, the increase in lake waterδ18O relative to in-flow water can confidently be linked to evaporative effects, much like the isotopic evolution of a cave drip-water that has fallen into a well ventilated cavern with low relative humidity [Lachniet, 2009]. The 13C-excess approach uses the isotopic co-variation that is a consequence of the combined effects described above to correct for the effects of in-flow water modification during lake residence. Through this process, the13C-excess approach serves, at least in part, to justify the critical assumption that lake carbonates accurately reflect the oxygen isotope composition of altitude sensitive regional meteoric water.
 In response to the second question, the strength and transferability of the co-variation slope value,m= 1, requires rigorous testing through both field-based research and laboratory experimentation. We stress that the co-variation slope value we report is based on a very limited data set, and should be extensively tested. For example, results from a previously published suite of evaporation experiments, conducted under elevatedpCO2conditions and variable calcite saturation states, demonstrate that carbon and oxygen isotopic compositions of calcite precipitates define a positive co-variant linear trend of slope of 2.2 [Ufnar et al., 2008]. However, the only experiment (experiment 9) that did not maintain its initially elevated pCO2 during the course of calcite precipitation, due to lost reaction vessel pressurization, showed a slope of ∼1 [Ufnar et al., 2008]. To better evaluate the strength and transferability of the co-variation slope we report, more studies of carbon and oxygen isotope compositions in fluvio-lacustrine systems spanning a broad range of physiographic and climatic contexts, in addition to further evaporation experiments modeling a variety of environmental conditions, are essential.
 Like all paleoelevation proxies, the 13C-excess approach has its strengths and weaknesses. The potential to distinguish topographic from climatic effects is a major advance as surface water evaporation can impose large, and potentially unrecognizable, uncertainties on paleoelevational interpretations when based on eitherδ18O or δD alone, as we have shown above. However, the 13C-excess approach will also yield inaccurate results in certain circumstances. For example, the isotopic composition of lake DIC in active geothermal areas may be sensitive to inputs of excess CO2 from magmatic sources (Figure 6). When large, magmatic carbon inputs can cause both isotopic disequilibrium effects associated with rapid CO2 degassing, in addition to modifications of δ13C-DIC values through equilibrium exchange reactions.
 Like all lake-mineral isotopic proxies, the13C-excess approach can lead to inaccurate interpretations in systems with large catchment areas. For example, the mean regional hypsometric altitude of Lake Cahuilla (southern California) is ∼500 m, yet the13C-excess proxy would yield an altitude of ∼2,000 m (Figure 6). Single-element (i.e.,δ18O or δD) lake-mineral elevation proxies would yield similarly inaccurate results as Lake Cahuilla in-flow is derived from the Colorado River catchment (mean hypsometric altitude ∼1,700 m) [Li et al., 2008]. In this example, isotopic proxies overestimate the immediately surrounding hypsometric mean altitude by >1 km, but accurately reflect the mean hypsometric altitude of the far-field in-flow water source regions in the Rocky Mountains and Colorado Plateau. Regardless of the isotopic proxy method being used, complementary provenance, paleocurrent, petrologic, textural, structural, thermochronologic, thermometric, and basinal analyses will improve the strength and accuracy of paleoelevational interpretations as they facilitate a deeper understanding of the regional geological and hydrological context through time.
5. Application to Cenozoic Lake Carbonates
 Published paleotopographic interpretations of a composite Neogene Mojave Desert record suggest regional surface lowering caused an ∼8‰ increase in lake carbonate δ18O values between the early Miocene and Pliocene [Horton and Chamberlain, 2006]. Molnar  has argued that this isotopic response may instead result from changes in regional climate, rather than elevation. The normalized δ18O and δ13C values of this record suggest that Molnar is likely correct: changes in regional surface elevations are not required to explain the large post-Oligocene increase inδ18O values (Figure 7). Specifically, the 13C-excess results suggest that a progressive shift toward drier climate conditions with longer lake water residence times and associated increases in evaporative modification of in-flow waters occurred, rather than a significant km-scale lowering of regional mean altitude. The relatively constant Neogene elevations of ∼1.7 km, as suggested by the13C-excess proxy, are similar to the modern regional mean hypsometric altitude of ∼1.5 km in the Death Valley region (Figure 7). In this example, the 13C-excess approach suggests previousδ18O-based paleotopographic interpretations byHorton and Chamberlain  are incorrect.
 A multiproxy isotopic record for the long-lived Elko basin has shown that the Sierra Nevada or present-day Great Basin, or both, uplifted ∼2 km between the middle Eocene and Miocene, followed by ∼1–2 km of surface downdrop between the middle Miocene and Pliocene [Horton et al., 2004]. The 13C-excess approach supports this interpretation, but also provides important estimates of the paleoelevation of the Cretaceous-Paleogene ‘Nevadaplano’ [Chamberlain et al., 2012]. Specifically, the 13C-excess approach demonstrates that regional elevations in the Paleogene were on the order of 1.5 to 2 km, in good agreement with published paleofloral estimates of Paleogene altitudes in northeast Nevada [Wolfe, 1994], but by the middle Miocene the region had uplifted a further ∼2–3 km, as predicted by the ‘SWEEP’ hypothesis [Chamberlain et al., 2012].
 The Elko record further indicates that regional surface lowering of at least 1 km occurred between the middle Miocene and Pliocene [Horton et al., 2004], different than the interpretation of the isotopically similar Mojave Desert record [Horton and Chamberlain, 2006]. These interpretations would require that evaporative modification was more severe in Neogene lakes in the Mojave than elsewhere in the Great Basin, an interpretation that is also supported by similar-age clay mineralδD and δ18O values from the same areas [Horton and Chamberlain, 2006]. Thus, despite both showing Neogene increases in δ18O, the 13C-excess approach suggests that the Mojaveδ18O record is best explained by a climatic shift to progressively more arid conditions, whereas the Elko record is best explained by surface lowering of the northern Sierra Nevada or the ‘SWEEP’ plateau, or both, since the middle Miocene. Furthermore, the ∼3 km altitude indicated by the Elko record during the Pliocene (Figure 7) is a reasonable quantification of the mean elevation of modern drainage divides upwind of Elko (rather than local elevations) suggesting little elevation change in these topographic barriers has occurred since the late Miocene as discussed by Mulch et al.  and Poage and Chamberlain . Application of the 13C-excess approach to the Elko record both supports prior multiproxy paleoelevation interpretations [Horton et al., 2004] and facilitates deeper understandings of the spatial and temporal interplay of topographic and climatic changes across the western U.S. cordillera that are not possible with single-element isotopic proxy methods.
 In a third example, Neogene lake carbonate δ18O records from the Lake Mead area suggest this region experienced ∼1.5 km of surface uplift during the Miocene [Horton and Chamberlain, 2006]. A strict paleoelevational interpretation of 13C-excess results supports this interpretation (Figure 7). However, we argue that the 13C-excess results more likely reflect Miocene changes in lake in-flow water source regions, analogous to the Quaternary Lake Cahuilla-Colorado River example described above (Figure 6). This alternative interpretation is supported by: 1) the similarity between the pre-middle Miocene Mojave and Lake Mead altitudes (Figure 7); 2) documented Miocene drainage pattern reorganizations in the region [Wernicke, 2011]; 3) the contemporaneous denudation of an adjacent continental plateau(s) [Flowers et al., 2008]. In this example, interpretation of 13C-excess results provides important physiographic detail that single-element isotopic elevation proxies fail to capture. Additionally, the13C-excess approach demonstrates how the initialδ18O-based paleotopographic interpretations byHorton and Chamberlain  are likely incorrect.
 Stable isotopic paleoelevation methods assume that the isotopic compositions of authigenic minerals are robust proxies of ancient meteoric waters. Our analysis of Quaternary lake carbonate stable isotopic records demonstrate that this assumption is not justified, due primarily to the effects of lake water evaporation. Lake carbonate δ18O values are systematically shifted toward more positive values than carbonates formed in equilibrium with meteoric-derived waters by several permil. As the water balance in ancient lake systems will almost certainly have changed through time, lake mineral isotopic compositions must be corrected for the effects of evaporation to be useful as paleoelevation proxies.
 The demonstration that lake carbonate δ13C values are also driven toward more positive values by evaporation (and associated conditions), and the δ13C versus δ18O positive co-variant trend that results, present an opportunity to correct for the effects of evaporation on lake mineral stable isotopic compositions using y-intercept values calculated from an empirically derived co-variation equation linking lake waters to in-flow waters, here defined as13C-excess values. Our analysis demonstrates that13C-excess values, excluding Tibetan lakes, accurately reflect Quaternary altitudes with a standard error of ∼500 m, in contrast to the km-scale errors traditional single-element paleoelevation proxies would yield for the same lake systems. Yet,13C-excess values must be interpreted with extreme caution in systems with large catchment areas or the potential for significant magmatic carbon inputs, and the slope of the line equation used to calculate13C-excess values requires further empirical, experimental, and theoretical study. Based on our analysis of carbonate isotopic compositions from widely distributed lakes, the13C-excess approach is a promising dual isotope-ratio method for estimating paleoaltitudes for ancient lake systems that is less sensitive to hydrospheric changes induced by long-term climate change than established single-element proxies.
 We thank Carmala Garzione, Michael Hren, Jay Quade, and an anonymous reviewer for their constructive reviews of earlier versions of this manuscript.