Past and present variability in leaf length of evergreen members of Nothofagus subgenus Lophozonia related to ecology and population dynamics

The lengths of fossil leaves of Nothofagus cunninghamii, and of closely related fossil species from sediments older than the Middle Pleistocene showed greater variation than fossil leaves from younger sediments and extant leaves from modern sediment samples and forest floor litter. Leaf sizes of samples of the other two extant evergreen species of Nothofagus subgenus Lophozonia, N. moorei and N. menziesii, were no more variable than modern N. cunninghamii. Leaf lengths of modern and Middle-Late Pleistocene N. cunninghamii and of modern N. menziesii were more or less unimodal in distribution, whereas the leaves of N. moorei follow bimodal distributions, and leaves from the Early Pleistocene Regatta Point and the Oligocene Little Rapid River sediments tended to be bimodal and highly variable. Leaf lengths from the Oligo-Miocene Monpeelyata sediments were also highly variable, but the sample size was too small to determine if they follow bimodal distributions. The bimodality and high variability in the older sites are unlikely to have been due to taphonomic processes. Thus, within population variability appears to have declined during the Late Tertiary-Early Pleistocene, although the presence of cryptic species in the fossil record could also explain the results. Both a decline in variability within species, and the extinction of Nothofagus species, are consistent with a well documented Late Cainozoic decline in rainforest diversity in Tasmania. Log transformed leaf lengths of modern forest floor litter samples of N. cunninghamii were strongly correlated with climatic parameters, particularly summer temperature, suggesting that this parameter is a strong determinant of leaf size in N. cunninghamii. Changes in N. cunninghamii leaf lengths within Middle Pleistocene sediments were associated with floristic changes interpreted as transitions in glacial/interglacial cycles, but these changes were small compared with changes predicted from climatic correlation of extant samples, probably because of limited genetic variability in local populations at the time of deposition.