DISCUSSION

This study explored a new concept to explain observed patterns in marine biogeography during the past 57 m.y., that a proxy for the diversity of habitats is a strong predictor of the diversity of macrofossils within a grid cell. However its pertinence should be tempered by at least two considerations. First, the geographic distribution and physical size of fossils used in this study limit extrapolation of these results. These data are dominated by marine bivalves and brachiopods, which although useful as a model for macroevolutionary studies, are subject to bias themselves stemming from several sources including shell preservation, small size, life habit, narrow geographic extent (Valentine et al. 2006).

Second, it should be acknowledged that although we developed a rigorous protocol the relationship between diversity and habitat type is particularly sensitive to the quality of the PD's lithological annotations (Peterson and Nakazawa 2008) and their translation from categorical to scalar quantities (see Appendix Tables 1, 2). It seems likely that the conversion of categorical data to mean scalar values (e.g., siltstone to 0% sand, 83.3% silt and 16.7% clay) may be too general to capture some subtler details necessary to distinguish, for example, siltstone deposition that occurs in different environments.

The analysis of differences in diversity measures across various tectonic settings presents an important test for our understanding of long-term biodiversity drivers. When taken over the entire period from 57 Ma to present, grid cells from eastern North America were the most diverse (Figure 3.1) despite being far removed from tectonic activity, while grid cells from tectonically active areas including Europe and Japan showed diversities that were only a fraction of that in eastern North America. Any hypothesis connecting the macrodiversity of a region to its tectonism should also account for these elevated diversities in passive regions.

On the other hand if considering only the tectonically active regions, tectonism and increased diversity did coincide with Europe and Japan where alpha diversity peaks during or near epochs of noted tectonic activity (Burbank et al. 1992; Takahashi and Saito 1997; Hall 2002, Vergés et al. 2002). Japan's grid cell diversity peaked during the Miocene, a phase of probable accelerated plate motion in the area during a continuum of tectonism there. Europe's grid cell diversity peaked during the Eocene, which also was a period of tectonic activity.

When both passive and active tectonic regions are both considered the variability in the number of habitat types, as determined through the diversity of lithologies present within a grid cell correlates reasonably to variability in diversity (species level: r = 0.763). We should, however, note that other relationships have a strong bearing on the overall biodiversity of an area. Particularly well observed are latitudinal gradients in biodiversity. We limited some latitudinal influence by constraining our analysis to regions of similar latitudinal zones. Our regions of eastern North America, Europe, and Japan are located between 25° and 60° north latitude.

We also considered that sampling resolution (grid cell size), such as was examined by MacNally et al. (2004), could alter the strength of the habitat type-alpha diversity relationship studied here. Sufficiently large grid cells would contain homogenized communities with low beta diversities while much smaller communities would appear more heterogeneous with higher beta values. We recognized the trade-off between grid cell size and geographic uncertainty, and our choice of a 1° by 1° resolution was based on the limitations of geographic resolution of coordinates from the PD. Tests using a data set with greater geographical resolution than was used here, could vary the grid cell sizes and may reveal a decreasing habitat type-alpha diversity r value with decreasing grid cell size.

In summary, the results of our study of variability in the diversity of macrofossils in passive and active tectonic settings are consistent with ecological niche theory. Tectonic plate movements are complex and can be the bearers not only of biodiversity reduction, but also biodiversification through the creation of new colonizable ecospace. Mild disturbance such as periodic eruptions or forest fires would provide for the often observed effect of increasing biodiversity, as is observed in successional forests subject to wildfire. Such disturbance events in eastern North America might have influenced biodiversity there just as effectively as volcanic disturbance events during the Eocene in the Mediterranean. An important challenge for any biodiversity-tectonic hypothesis should be to explain why tectonic disturbances should be better agents of biodiversification than disturbance events in passive areas. Our results indicate that drivers of biodiversity in passive tectonic environments are at least as effective as biodiversity drivers in convergent tectonic settings.

Our data are based on 83,213 fossil occurrences that were partitioned into 1,565 spatially equivalent grid cells, which spanned the most recent 57 m.y. These data indicate that the most robust predictor of alpha diversity (at genus or species level) in both passive and active tectonic settings is the number of habitat types which is consistent with modern ecological observations where diversification occurs as a response to limitations, and species tend to specialize and diversify as they partition themselves along resource gradients.