| Co-Authors: | Upchurch, Garland R., and Knaus, Margaret J., Southwest Texas State University, gu01@swt.edu; Mack, Greg H., New Mexico State University |
| Abstract: | The statistical relationship between leaf physiognomy and climate for modern plants is widely used by paleobotanists to reconstruct Late Cretaceous climate. Methods such as simple linear regression (SLR) and multiple regression (MR) are used to derive leaf/climate relationships for modern floras, which then are used to estimate temperature and precipitation from paleofloras. When applied to modern test sites, physiognomic methods provide good "ballpark estimates" of mean annual temperature (MAT) and mean annual precipitation (MAP) but can show significant discrepancies between estimated and measured climate. The magnitude of the discrepancy (up to 5°C for MAT) is determined by the choice of equation used to quantify the leaf/climate relationship, referred to here as the physiognomic transfer function.
To determine the extent to which estimates of Cretaceous climate are biased by the choice of physiognomic transfer function, we applied a variety of published physiognomic transfer functions to a well-preserved leaf megaflora from the latest Cretaceous (Maastrichtian) Jose Creek Member of the McRae Formation, southern New Mexico. The Jose Creek assemblage represents probable in situ leaf litter preserved in volcanic ash that was derived from vegetation growing in well-drained soil, providing sampling conditions close to those represented by many modern reference floras. The analyzed assemblage consists of 42 species of dicot leaves, which co-occur with 10 species of ferns, conifers, and thermophilic monocots. Calculations of MAT and MAP were made using raw physiognomic data and data adjusted to account for the over-representation of small leaves in modern leaf litter.
For the Jose Creek assemblage, calculated MAT is 16-25°C and calculated MAP is 600-1200 mm/yr, a range of over 8°C for MAT and 600 mm/year for MAP. For temperature, SLR models give less variable results than MR models. This is curious, because each SLR model of temperature is derived from a geographically distinct reference flora, while all MR models of temperature are derived from a reference flora that comprises different versions of the CLAMP (Climate Leaf Analysis Multivariate Package) database. For precipitation, each SLR and MR model provides somewhat variable results, depending on whether leaf size values are adjusted for taphonomic bias towards the preservation of small leaves. Variation in calculated temperature and precipitation, relative to the mean of all estimates, can exceed + 35%. However, the range of values is congruent with more qualitative estimates of climate derived from paleosols and plant life form, which indicate a warm subhumid subtropical climate for the Jose Creek Member. This implies that physiognomic transfer functions are probably accurate to the level of climate zone. However, further testing of physiognomic transfer functions is needed to determine their reliability and level of accuracy. |