Biology Reference
In-Depth Information
the armadillo in North America, for example, would be quite different from
one based on its present distribution.
The possibility of changes in ecological requirements over time and the
incomplete range of some extant plants is widely recognized, and there are
ways of mitigating both limitations. One is to base paleoenvironmental
reconstructions on assemblages, rather than on the ecological and habitat
characteristics of a single or limited number of “key,” or “indicator,” plants.
The other is to consider all the fossil fl oras and faunas in the region to give
a fuller picture of the biotic history, in other words, a context. It would be
singularly improbable that all organisms from multiple assemblages would
change their requirements at the same time in the same direction.
Another way of reconstructing environments, in addition to the NLR or
modern analog method, is from the external morphology of leaves, or leaf
physiognomy. The value of the approach is that it is independent of the NLR
method, and it does not depend on identifi cation of the fossils. Early study
of extant plants showed that some leaf features correlated with rainfall and
to a lesser extent with temperature. Bailey and Sinnott (1915) provided the
following tabulation:
% entire
% nonentire
Cold temperate environment (eastern U.S.)
Trees
10
90
Shrubs
14
86
Woody
13
87
Warm tropical environment (Amazonia)
Trees
90
10
Shrubs
87
13
Woody
88
12
Later studies showed that larger leaves, thinner texture, and drip tips
occur frequently in tropical plants, while thicker, coriaceous, and smaller
leaves are found most commonly in plants of dry habitats. Wolfe (1990,
1993) tabulated these features for several modern vegetation types in differ-
ent parts of the world, and developed a computer program called CLAMP
(Climate-Leaf Analysis Multivariate Program). Using this program, data on
leaf morphology from fossil fl oras can be entered, generating an estimate of
the paleoclimate (usually erring slightly on the cool side). The results can
be cross-checked with those from the NLR method, and collectively they
