Leaf physiognomic indications for a moist warm-temperate climate in NW Italy during the Messinian (Late Miocene)

Abstract

A leaf physiognomic analysis has been carried out on 980 specimens of Messinian leaves collected during the last three decades of the 20th century from the Gessoso Solfifera Formation (Messinian, Late Miocene) in NW Italy. The material is stored at the F. Eusebio Museum of Alba, and originates from a few sites located in close proximity to each other in the valley of the Tanaro River near Alba. According to recently published age models, the studied plant-bearing sites can be dated between 6.0 and 5.7 Ma. The physiognomic analysis was preceded by a taxonomic one, which allowed separation of 74 to 77 dicotyledonous taxa (morphotypes). The botanical affinity of a large proportion of the taxa is not yet defined, because they either belong to the so-called “leguminosoid” types or consist of entire-margined leaves with a coriaceous texture, whose affiliations with a known plant family are not definable solely on the basis of gross morphology. Several recently published univariate and multivariate correlations between climate and leaf physiognomy were applied to the individual floras in order to estimate selected palaeoclimatic parameters. The results obtained for the two richest assemblages, as well as the combined flora from all localities, point to warm-temperate and humid palaeoclimatic conditions, also indicated by the taxonomic composition of these floras. One other assemblage, which consists of fewer taxa, provides significantly lower temperature and precipitation estimates. However, in this case, it is rather likely that the estimates have been influenced by the low number of taxa, caused by taphonomic biases, reducing the representation of entire-margined taxa. Our results imply that at least some existing leaf physiognomic techniques can reliably be used for the reconstruction of palaeoclimatic conditions based on floras from Europe, although previous data from other European floras suggested that some of these methods may produce erroneous results.