Trilobite diversity and the Ordovician Radiation: a reply to Miller et al.

    The comments by Miller et al. boil down to the questions of the scope of our data set and its relevance to community reorganization during the Ordovician.  Although we were constrained by our desire to include only high-quality, quantitative data in our analysis, we feel that we have produced a comprehensive survey of continental scale.  It is true that our Upper Ordovician data are from exclusively Canadian sites, but in discussing our work, Miller et al. treat Canada as if it were the size of Rhode Island or Delaware.  Canada is actually larger than the continental United States and our Upper Ordovician study sites are spread across the country: southern Ontario, Manitoba, the Arctic Archipelago, and the Mackenzie Mountains of the northern Cordillera.  Lines connecting these regions circumscribe an area of more than 3,000,000 km2.  As even a casual glance at our appendix will show, our data are not confined to shallow-water carbonates, but include a number of collections from subtidal shales and storm deposits (including cool-water facies from the Trenton Group [Brookfield 1988]) as well as from a variety of nearshore, buildup, and deep subtidal settings.  It is also worth pointing out that our Middle Ordovician data are more extensive than those from the Upper Ordovician and yet still show levels of alpha diversity that are comparable to the Late Cambrian, a result that is not predicted by alternative models of trilobite history during the Ordovician Radiation (Sepkoski and Sheehan 1983; Sepkoski and Miller 1985).
    Miller et al.'s comments are all the more perplexing when one compares the scale of our observations with those that are marshaled to counter them.  The Cincinnati Arch and Nashville Dome areas are separated by a N-S distance of only about 250 km and a published estimate of the total area of the basin is only 250,000 km2 (Patzkowsky and Holland 1997).  In Droser's work (Droser et al. 1996), the Ordovician Radiation is apparently captured adequately in the Ibex region of Utah, an area that is orders of magnitude smaller than our geographic coverage.
    We make no apologies for having restricted our analysis to Laurentia.  Laurentian data have been the foundation of contemporary efforts to characterize the Ordovician radiations (Sepkoski and Sheehan 1983; Sepkoski and Miller 1985; Droser et al. 1996) and it is a perfectly reasonable place to start.  We have expanded our work geographically to other parts of the world and temporally into the Silurian and, while the results will be presented fully elsewhere (Adrain, Westrop, and B. D. E. Chatterton unpublished data), some of our conclusions are relevant to this discussion.
    Table 1 presents data for Middle and Upper Ordovician shallow and deep subtidal facies of Laurentia (data set expanded from that used by Westrop and Adrain [1998]), Baltica, and Avalonia, with smaller amounts of information for Australia, Armorica, and various Chinese terranes united under "other provinces."  Much of the published information for extra-Laurentian trilobite faunas lacks the detailed abundance data that we assembled for Laurentia, but a subset can be rarefied and we have tabulated expected species number [E(Sn)] at a standard sample size of 90 individuals.  In both raw species counts and rarefied data, mean values for Laurentian alpha diversity are in line with values obtained from other biogeographic regions; indeed, highest mean values for shallow subtidal species richness occur outside Laurentia.
    In this context, species richness data tabulated by Miller et al. (Tables 1 and 2) for the Cincinnati region and the Nashville Dome of central Tennessee are decidedly low and are certainly not representative of either Laurentian or global patterns.  The depauperate nature of the data for shallow subtidal facies of the Nashville Dome is particularly striking.  Indeed, one would be hard pressed to find another Middle Ordovician shallow subtidal sequence anywhere else in North America with as few as three or four trilobite species.  For example, Middle Ordovician shallow subtidal collections from eastern Tennessee in our data set (Benbolt Formation at Cedar Springs and Evans Ferry [erroneously listed as from Virginia in our appendix]) contain more than double the number of species.  Similarly, slabs from a single horizon in the Middle Ordovician Bromide Formation of Oklahoma (Westrop unpublished data) yield ten species [n = 95; E(Sn) = 9.84) in a surface area of only 0.5m2.  Other Middle Ordovician shallow subtidal collections in our data set from Ontario (Trenton Group, Verulam Formation), the Upper Mississippi Valley (Platteville Group), the Mackenzie Mountains (Sunblood and Esbataottine Formations), and the Laurentian terrane of the Girvan area, Scotland (Stinchar Limestone), tell the same story.  Moreover, in the Silurian, mean values of E(Sn) for Llandovery (9.73) and Wenlock (8.6) and Ludlow (8.83) shallow subtidal trilobite collections from Laurentia (Adrain, Westrop, and B. D. E. Chatterton unpublished data) are also substantially higher than those reported from the Nashville Dome.
    The compilation for the Nashville Dome is said to be based on "fieldwork and published literature."  And yet, modern systematic studies of Ordovician trilobites do not exist for this area.  Indeed, one has to go back  as far as works such as Bassler (1932) just to find faunal lists.   There has been a long history of collecting in the Cincinnati region but, even here, a revision of the trilobite fauna is long overdue. Thus, lack of modern systematic treatment alone may explain some of the disparity between Miller et al.'s data and the global picture.  Until trained systematists undertake a comprehensive field-based revision of the faunas in both areas, their relevance to questions of Ordovician trilobite diversity remains uncertain.
    Our data indicate that trilobite alpha diversity was maintained at levels comparable to the Cambrian in  Middle to Upper Ordovician shelf environments of Laurentia and that these values are paralleled in other biogeographic regions.  In our experience, the only Laurentian shelf facies in which trilobites assemblages are of low diversity (typically fewer than four species) are those of nearshore settings and, especially, in the Taconic clastic wedges associated with infilling of the Appalachian basin (flysch of Lehmann et al. 1995).  The fact that trilobites are locally abundant in shales in the Cincinnati Arch region does not falsify the latter observation.  These occurrences of trilobites in shales are limited to a few stratigraphic intervals, each of which is no more than a meter in thickness (Frey 1987; Schumacher and Shrake 1997), that can be traced over parts of southern Ohio and eastern Indiana.  Although two trilobite genera (Isotelus and Flexicalymene) may represent up to about 25% of specimens in these mollusc-dominated paleocommunities (Frey 1987), we think that the complete absence of a variety of other trilobite taxa that account for most of the species diversity in typical Middle and Upper Ordovician shallow subtidal biofacies (e.g., illaenids, cheirurids, encrinurids, pterygometopids, and lichids [see, for example, Ludvigsen 1978]) is far more significant and supports our conclusions.  The "trilobite shales" certainly attest to the fact that a small subset of species could be locally abundant in Laurentian clastic facies, but they tell us little about Ordovician trilobite paleoecology in general.
    The shell bed study by Droser and colleagues (Li and Droser, in Droser and Sheehan 1995: p. 88; Droser et al. 1996) is not directly relevant to the question of trilobite diversity because their data document only relative abundances of classes or phyla.  This provides a record of community change at the broadest scale, but it is hardly an unbiased archive.  We do not contest a decline in relative abundances of trilobites but we hesitate to infer the drop in absolute abundances implied by Droser et al.  As we argued in our paper, the same pattern of apparent decline will result simply from the great expansion of community membership during the Ordovician Radiation.  Moreover, because of the effects of taphonomic factors, a literal interpretation of the shell bed record is at best hazardous.  Clearly, we did not mean to imply that trilobites could not form shell beds-our own work indicates otherwise-but Miller et al. miss the point.  The paper that they cite (Westrop 1986) demonstrated that size- and/or shape-sorting in Cambrian shell beds can produce dramatically different relative abundance profiles for trilobite assemblages drawn from the same biofacies.  Similar taphonomic obstacles to interpretation have been documented in Cretaceous non-marine microvertebrate assemblages (Blob and Forillo 1996) and we find it hard to believe that Ordovician shell beds would be immune from these types of problems.  The use of shell bed types is certainly a novel approach to documenting the changing composition of Ordovician faunas, but the very nature of the data demands utmost caution when drawing conclusions.
    Taphonomic overprint may make direct estimation of the absolute abundances of trilobites an intractable problem.  However, there may be indirect approaches.  If we are correct in our assessment of alpha diversity, then any significant real decline in trilobite abundances in local habitats would have occurred in the face of essentially constant species richness.   This would translate into a drop in mean population sizes which, in turn, should be expressed by an increase in species turnover rates and a reduction in species longevity.  An accurate compilation of species ranges of even Laurentian trilobites would be a major undertaking and is clearly beyond the scope of this note.  However, Foote's (1988) analysis of generic survivorship indicates that no change in turnover rates took place during the Middle and Upper Ordovician.  Rather, the pivotal point in the survivorship patterns lies between the Cambrian and Ordovician, with the lower Tremadoc being somewhat transitional.  As recognized by Foote (1988: p. 268), the increase in genus survivorship and longevity between the Cambrian and Ordovician has a variety of potential explanations, including differences in traditional approaches to systematics.  However,  genus level turnover in the Middle and Upper Ordovician is not likely to be influenced by taxonomic bias and appears to be as stable as our estimates of alpha diversity.
    In a final attempt to dismiss our conclusions, Miller et al. suggest that we may simply have failed to sample those localities in which the Ordovician Radiation was "going full tilt."   Here, we believe that they are being disingenuous.  They contrast our work with the earlier study by Westrop et al. (1995) on Cambrian-Ordovician nearshore paleocommunities, without noticing that collections and localities used in that study are incorporated into our present analysis.  Moreover, Miller (Sepkoski and Miller 1985) used data from many of the same regions in his characterization of the Ordovician reorganization of communities.  For example, we both included collections from the Middle Ordovician of the Mackenzie Mountains (communities 61-65 in Sepkoski and Miller: p. 183) and yet Sepkoski and Miller had no doubt that the Ordovician Radiation was "going full tilt" in that area.  The Ordovician Radiation was also alive and well in Middle Ordovician of western New York, a major source of information in Sepkoski and Miller's study (1985: p. 183, communities 67-70, 75, 79, 80-81, 83-84), but Miller et al. are apparently unwilling to accept the relevance of data from the same facies and faunas some 200 km to the west in south-central Ontario.  In  Manitoba, the diverse trilobite fauna of the Red River Formation is part of a paleocommunity dominated by tabulate and rugose corals, stromatoporoids, receptaculitids, and cephalopods (Westrop and Ludvigsen 1983), and the same paleocommunity type (the "Arctic Ordovician Fauna") is associated with the equally rich trilobite assemblages of the Arctic Archipelago (Thorsteinssen 1958; Bolton et al. 1977).  If the "Arctic Ordovician Fauna" is an unequivocal product of the Ordovician Radiation in Nevada (Sheehan, in Sheehan and Droser 1995: p. 83), then it is surely of the same origin in Canada. There is no reason to believe that our data are anything but an accurate reflection of the diversity history of trilobites in habitats that record the Ordovician Radiation in Laurentia.
    There is a broad consensus that trilobites became reduced in relative importance during the Ordovician Radiation, but we are still far from understanding how this was accomplished.  Our conclusions about the stability of Laurentian trilobite alpha diversity are novel and run counter to conventional wisdom.  The available data indicate that within-habitat species richness was equally stable in other biogeographic regions.  Rather than struggling to dismiss these results, we suggest that a more fruitful approach is to use them to constrain hypotheses about the dynamics of the radiation.  They indicate that community expansion, rather than displacement of one evolutionary fauna by another, is a dominant theme.  They also suggest that controls on taxonomic diversity of trilobites during the Ordovician are lodged at scales beyond those regulating species richness in local habitats.  In a new compilation of global taxonomic richness (Adrain et al. 1998), we have corroborated a general Ordovician decline in the number of trilobite genera.  Even here, the pattern is more complex than previously appreciated, with a subset of taxa, the Whiterock Fauna, undergoing a dramatic Middle Ordovician radiation alongside members of the Paleozoic Evolutionary Fauna.  The apparent decoupling of local species diversity from global taxonomic diversity suggests that factors such as declining provinciality during the Ordovician (Whittington and Hughes 1972) may be important in the history of the group.  By taking a fresh look at trilobite diversity at a variety of scales, it may be possible to gain new perspectives on the history of community reorganization during the Ordovician.

References
Adrain, J. M., R. A. Fortey, and S. R. Westrop.  1998.  Post-Cambrian trilobite diversity and Evolutionary Faunas.  Science.
Bassler, R. S.  1932.  The stratigraphy of the Central Basin of Tennessee.  Tennesse Division of Geology Bulletin 38.
Blob, R. W., and A. R. Fiorillo.  1996.  The significance of vertebrate microfossil size and shape distributions for faunal
    abundance reconstructions: a Late Cretaceous example.  Paleobiology 22:422-435.
Bolton, T., B. Sanford, M. Copeland, C. Barnes, and J. K. Rigby. 1977. Geology of Ordovician rocks, Melville Peninsula and
    region, southeastern District of Franklin. Geological Survey of Canada, Bulletin 269.
Brookfield, M. E., 1988.  A mid-Ordovician temperate carbonate shelf-the Black River and Trenton groups of southern Ontario.
    Sedimentary Geology 60:137-153.
Droser, M. L.,  R. A. Fortey, and X. Li .  1996.  The Ordovician Radiation.  American Scientist 84:122-131.
Droser, M. L., and Sheehan, P. M.  1995.  Paleoecology of the Ordovician radiation and the Late Ordovician extinction event:
    evidence from the Great Basin.  Pp. 64-106 in Ordovician of the Great Basin: fieldtrip guidebook and volume for the Seventh
    International Symposium on the Ordovician System. Pacific Section, SEPM, Fullerton, Calif..
Foote, M., 1988.  Survivorship analysis of Cambrian and Ordovician trilobites.  Paleobiology 14:258-271.
Frey, R. C.  1987.  The paleoecology of a Late Ordovician shale unit from southwestern Ohio and southeastern Indiana.  Journal
    of Paleontology 61:242-267.
Lehmann, D., C. E. Brett, R. Cole, and G. Baird.  1995.  Distal sedimentation in a peripheral foreland basin: Ordovician black
    shales and associated flysch of the western Taconic foreland basin, New York State and Ontario.  Geological Society of
    America Bulletin 107:708-724.
Ludvigsen, R.  1978.  Middle Ordovician trilobite biofacies, southern Mackenzie Mountains.   Pp 1-37 in  C. R Stelck and B. D.
    E. Chatterton, eds.  Western and arctic Canadian biostratigraphy.   Geological Association of Canada Special Paper 18.
    Patzkowsky, M. E., and S. M. Holland.  1997.  Patterns of turnover in Middle and Upper Ordovician brachiopods in the
    eastern United States: a test of coordinated stasis.  Paleobiology 23:420-443.
Sepkoski, J. J., Jr., and A. I. Miller. 1985.  Evolutionary faunas and the distribution of Paleozoic benthic communities in space and
    time.  Pp. 153-190  in  J. W. Valentine, ed.  Phanerozoic diversity patterns: profiles in macroevolution: Princeton University
    Press, Princeton, N.J..
Sepkoski, J. J., Jr., and P. M. Sheehan. 1983.   Diversity, faunal change and community replacement during the Ordovician
    radiations.  Pp. 673-717 in  M. J. S. Tevesz and P. M. McCall, eds. Biotic interactions in Recent and fossil benthic
    communities. Plenum, New York.
Shumacher, G. A., and G. L. Shrake.  1997.  Paleoecology and comparative taphonomy of and Isotelus (Trilobita) fossil
    laggerstätten from the Waynesville Formation (Upper Ordovician, Cincinnatian Series) of southwestern Ohio.  Pp. 131-161 in
    C. E. Brett and G. C. Baird, eds.  Paleontological events.  Stratigraphical, ecological and evolutionary implications.  Columbia
    University Press, New York.
Thorsteinsson, R. 1958.  Cornwallis and Little Cornwallis Islands, District of Franklin, Northwest Territories.  Geological Survey of
    Canada Memoir 294.

Stephen R. Westrop.  Oklahoma Museum of Natural History and School of Geology and Geophysics, University of Oklahoma, Norman, Oklahoma 73019. E-mail:swestrop@ou.edu

Jonathan M. Adrain.  Department of Palaeontology, The Natural History Museum, Cromwell Road, London SW7 5BD, United Kingdom. E-mail: j.adrain@nhm.ac.uk