DO PALEOECOLOGISTS HAVE TO CHOOSE BETWEEN ACCURATE TEMPORAL PLACEMENTS AND ACCURATE CLIMATIC VALUES WHEN EXAMINING STRATIGRAPHIC TRENDS?

Correlations between taxonomic diversity of modern mammals and climatic variables were established using datasets generated from ecoregion maps. The resulting predictive equations from the statistically significant correlations were used to estimate Pliocene temperature values and trends using the fossil mammals in the Glenns Ferry Formation of Hagerman Fossil Beds National Monument (HAFO), Idaho. The estimated climatic variables were then compared to independent temperature proxies.

Initial temperature trends based on the fossil mammals matched external data in general pattern, but the minimum temperature peaks were offset by about 200 ka. This is a huge discrepancy given the ~1.2 Ma temporal duration represented by the Glenns Ferry Formation at HAFO. The offset of the temperature patterns appear to be a result of real and artificial time-averaging factors. Initial temperature estimates for HAFO were generated using a sliding 20-m stratigraphic window; the size of the sliding window impacts both the estimated values and the overall pattern. Smaller sliding windows more accurately place the temperature minimum toward the top of the stratigraphic section; however, the chance of missing an included species, thereby creating different estimated temperature patterns, is increased. Different details in the patterns occur because several taxa at HAFO are only known from a few stratigraphically-scattered specimens. However, using a range-through approach, which considers a taxon present if it occurs both higher and lower stratigraphically, would exclude the possibility of extirpation.

At HAFO, some degree of time-averaging is necessary because few individual localities can be said to have a large proportion of the mammalian fauna present at that particular time. Time averaging is commonly known to remove the short term noise, but that is not seen consistently in this dataset. Applying different sizes of sliding windows to the ecoregion models developed, produced both different details and temporal shifts in the position of peak values. Further, the effects vary based on the particular model being applied. Rather than sacrificing accuracy in either temporal position or estimate temperature, paleoecologists can examine stratigraphic intervals with multiple size windows, and use a combination of best scenarios.