The early Paleogene represents the warmest time of the Cenozoic. Tropical temperatures may have been ≥35 °C (Head et al., 2009; Jaramillo et al., 2010; Pearson et al., 2001). Mean annual temperatures (MAT) in mid-latitude continental interiors were considerably higher than today, with average winter temperatures likely above freezing (e.g., Fricke and Wing, 2004; Greenwood and Wing, 1995; Markwick, 1994, 1998; Wing et al., 2005). Temperatures in the Arctic were significantly above freezing, despite prolonged darkness during winter (e.g., Eberle et al., 2010; Greenwood and Wing, 1995; Markwick, 1994, 1998; Sluijs et al., 2006), and oceanic bottom waters generated at high latitudes were ∼≥10 °C warmer than modern (e.g., Zachos et al., 2001). It has also often been argued that early Paleogene climate was equable, or in other words, the mean annual range of temperatures (MART) was relatively small and considerably less than today (e.g., Greenwood and Wing, 1995; Markwick, 1994, 1998; Wing and Greenwood, 1993).

Climate modeling studies of the early Paleogene have been able to generate MAT estimates for mid-latitude continental interiors in agreement with proxy data, but they are unable to reproduce high polar temperatures and mid-continental winter temperatures without also suggesting summer temperatures high enough to potentially negatively impact plant photosynthesis (i.e., ≥40 °C in middle to low latitudes) (e.g., Huber and Caballero, 2011; Sewall and Sloan, 2006; Sloan and Barron, 1990; Sloan et al., 2001; Winguth et al., 2010). This model-proxy discrepancy has often been referred to as the “equable climate problem” (reviewed by Huber and Caballero [2011]).

Although most arguments for equability come from terrestrial records (e.g., Greenwood and Wing, 1995), quantitative estimates of seasonality using terrestrial proxies are lacking. Evaluation of the frost-tolerance of crocodilians and plants, such as palms and cycads, has been used to argue for cold month mean temperatures (CMMT) above freezing (Greenwood and Wing, 1995; Markwick, 1994, 1998). Although these studies can provide minimum temperature boundaries (e.g., no temperatures below freezing), they cannot quantitatively estimate CMMT. Attempts have been made to quantitatively estimate seasonal temperatures using paleobotanical methods, but these are unreliable. Two of the most important problems with reconstructing seasonal temperatures are: (1) the methods and the data sets that have typically been used to estimate seasonality are fraught with problems (see a discussion of problems with the Climate Leaf Analysis Multivariate Program [CLAMP] in Peppe et al. [2010]); and (2) in the modern calibration data set used to develop the seasonality proxies, there is no relationship between leaf physiognomy and CMMT after accounting for the covariation of MAT (Jordan, 1997), making it impossible to reliably estimate seasonality. Further, no studies to date have been able to estimate warm month mean temperature (WMMT). Thus, it is unclear if the equable climate problem is the result of inability of climate models to accurately reproduce equability during greenhouse climates, or because climate proxies underestimate MART.

Resolving the equable climate problem is particularly important because it has considerable ramifications for understanding the dynamics of past and future greenhouse climates. If proxy data are correct, and past greenhouse climates were more equable than today’s, fundamental aspects of greenhouse climates must be missing from climate models. This would indicate that unknown biases exist in climate models related to missing forcing factors, because regional and global climate models cannot reproduce climate parameters that agree with proxy data while maintaining equable climates. On the other hand, if climate data generated by models are accurate and greenhouse climates were not more equable than modern, winter temperatures were much cooler than proxy data indicate, or summer temperatures in mid and low latitudes were warm enough to potentially negatively impact plant photosynthesis. This possibility would suggest that physiological responses of plants and animals to climate extremes were different in the past, or that terrestrial biota are more adaptable to extreme climates than thought. The equable climate problem thus shows clearly that aspects of greenhouse climates are poorly understood.

Snell et al. (2013, p. 55 in this issue of Geology) address the equable climate problem through a novel multiproxy approach that uses paleobotanically derived MAT estimates coupled with WMMT estimates from carbonate clumped isotope thermometry of pedogenic carbonates to estimate MART at mid-continental late Paleocene–early Eocene sites in the Bighorn Basin in Wyoming (United States). They conclude that early Paleogene and modern seasonal temperature variability are similar: the clumped isotope estimates for maximum summer temperatures are on average ∼30–35 °C; i.e., consistently 15–20 °C warmer than mean temperature estimates from fossil floras. A high-resolution regional model shows results corresponding with the proxy data. Snell et al. thus conclude that terrestrial proxies have consistently underestimated MART because of their inability to accurately estimate WMMT, and that climate models have consistently underestimated mean annual and seasonal temperatures, so that the equable climate problem is an artifact of biases in both climate models and proxy data.

If clumped isotope temperature estimates accurately reflect mean or maximum summer temperatures as suggested by modern calibration studies (Passey et al., 2010; Quade et al., 2011), Snell et al. indicate maximum summer temperatures of 30–40 °C, which has at least two interesting implications beyond demonstrating that climates were not equable during the early Paleogene. First, mid-latitude continental summer temperatures in the early Paleogene may have been higher than tropical temperatures today, and lower latitudes may have experienced even higher summer temperatures. Assuming a mild Eocene temperature gradient of ∼0.4 °C/°latitude, summer temperatures in low latitudes would have been ≥40 °C. Interestingly, these extreme temperatures at mid and low latitudes implied by the Snell et al. study are similar to continental summer temperature estimates of Eocene climate produced by global climate models assuming very high levels of CO2 (e.g., Huber and Caballero, 2011), suggesting some potential congruence between these new proxy results and global models.

Second, the Snell et al. summer temperature estimates for the Bighorn Basin, and those implied for low latitudes, indicate that summer temperatures would have been high enough to be detrimental to plants’ photosynthetic capacity (e.g., Berry and Bjorkman, 1980), suggesting that Paleogene plant communities in mid and low latitudes may have been commonly growing under physiological stress. This seems unlikely given their composition and diversity (e.g., Wilf, 2000; Wing et al., 2005). Alternatively, it is possible that early Paleogene plants were better adapted to very warm climates than modern plants, or that the high CO2 conditions at the time (e.g., Beerling and Royer, 2011) allowed plants to thrive in conditions that would be deleterious today. Growth-chamber experiments have shown that some plants are better able to adapt to extreme warmth (e.g., Hozain et al., 2010; Silim et al., 2010), and that increased photosynthesis related to high CO2 offsets a decline in photosynthetic capacity at high temperatures (Lloyd and Farquhar, 2008), suggesting that, under the right conditions, at least some plants can thrive at extreme temperatures. This is an exciting possibility and would help explain why there is no evidence for plant extinction in the tropics during the Paleocene–Eocene thermal maximum (Jaramillo et al., 2010).

The Snell et al. study provides a powerful new methodology for reconstructing seasonality in terrestrial settings through the combination of paleobotanically derived MAT estimates and clumped isotope estimates of summer temperatures. As noted by the authors, this study represents a single data point from a continental site in the early Paleogene, and clearly more work must be done to fully address the equable climate problem. Nonetheless, it offers the intriguing possibility that greenhouse climates were not more equable than modern climate. Further, it suggests that some fundamental aspects of plant communities’ ability to live in extremely warm climates may have been different in the past. This raises numerous questions about both the dynamics of warm climates and the response of terrestrial ecosystems in their adaptation to extreme temperatures, which need to be addressed by future modeling and proxy-based studies.