INTRODUCTION

The factors driving evolutionary change in biotas are controversial. Both primarily internally driven and externally controlled evolutionary change have been proposed. At one end of the spectrum, the Red Queen model posits primarily internal, biologic interaction driven evolution (van Valen 1973; Stenseth and Maynard-Smith 1984), while at the other extreme, Stenseth and Maynard-Smith (1984) stationary model, or Vrba's (1980, 1985) pulsed model posit that evolutionary change is dominantly driven by change in the external physical environment. Although numerous studies of biotic evolution have been published using macrofossil data, marine microfossil data offers, in principle, several advantages for this type of study, including relatively complete, species-level data sets, good chronology, and the widespread availability of independent, detailed paleoenvironmental information.

Marine micropaleontologists have only infrequently addressed this problem directly, being primarily concerned with research in biostratigraphy and paleoenvironmental reconstructions. Yet, at least in deep-sea plankton micropaleontology, these primary fields of inquiry suggest that both external and internal factors may be important. A close correlation between species distributions and external environmental factors has long been known and has been widely exploited by using fossil distributions as a proxy for past environmental parameters such as temperature (Imbrie and Kipp 1971; CLIMAP project members 1976; Kennett 1982). Yet in biostratigraphic work, it is generally assumed that species origins and extinctions occur more or less simultaneously throughout a broad geographic range (Bolli et al. 1985), despite the existence of environmental gradients in the oceans. Shifts in these gradients during the normal course of gradually changing global environments would cause critical absolute values of environmental thresholds to be crossed at noticeably different times in different locations. Biostratigraphic event isochroneity thus would seem to violate the basic assumption that a species' distribution, including the bounding value of zero abundance (local extinction), is closely controlled by specific combinations of external environmental parameters. Although detailed evaluation of biostratigraphic event isochroniety has shown this assumption sometimes to be wrong (Johnson and Nigrini 1985; Moore et al. 1993), there are also studies which have demonstrated that it is often correct (Hays and Shackleton 1976; Thierstein et al. 1977; Flores et al. 2000; Raffi, 2002).

Studies of macroevolutionary patterns in marine microfossil plankton do exist that have addressed the question of biotic vs abiotic controls on macroevolutionary change. Hoffman and Kitchell (1984) were among the first to examine macroevolutionary patterns with the explicit goal of testing biotic vs abiotic hypotheses. They studied several groups of Cenozoic marine plankton, but, due to the lack of comparisons to paleoenvironmental data, their results were inconclusive. Kitchell (1987) subsequently compared Hoffman and Kitchell's (1984) evolutionary patterns to the published literature of paleoceanographic change but was unable to detect significant correlations between evolution and environment. Most other work by contrast has demonstrated (or at least argued) for a strong link between environmental change and evolutionary change (e.g., Lipps 1970; Stehli et al. 1972; Hart 1980; Thunell 1981; Wei and Kennett 1983, 1986; Hallock 1987; Roth 1987; Stanley et al. 1988; Corfield and Shackleton 1988; Hallock et al. 1991; Pearson 1996). Several different environmental factors have been considered by these authors as possible pacesetters for evolutionary change, including changes in water temperature, global latitudinal gradients in the physical environment; water column stratification and productivity. In general, the results of these studies suggest that productivity and water column stratification play important roles in regulating rates of faunal change. Other more biologic or internal factors, such as the geographic extent of species, or the prior longevity of a species, by contrast, do not appear to play an important role. Nor has temperature been shown to be the most important direct factor. Differences in mean longevity and diversification history are, however, observed between individual clades within major taxonomic groups, suggesting that other, often poorly understood, biologic adaptations also play important roles.

The work done so far on macroevolutionary change in marine microplankton nonetheless still has some limitations. Most importantly, the vast majority of these studies are of one fossil group - the planktonic foraminifera - which, despite their enormous importance to paleoceanography and biostratigraphy research, are a relatively modest component of the modern plankton, measured either by taxonomic diversity (ca. 40 living morphospecies) or abundance in the plankton. Thus this group may not be fully representative of plankton organisms in general. Planktonic foraminifera are also primarily found in the 'warm water lid' of the ocean- (i.e., the low to mid latitude regions and only in the upper water column). This is significant, as within this environment many of the large environmental changes that have occurred elsewhere in the oceans over time have been muted. Most of the global temperature change signal in the Cenozoic, for example, has been concentrated in the higher latitude regions, as have the most dramatic changes in surface water-mass distribution and circulation, particularly the formation of the circumpolar Antarctic Current and the Southern Ocean in the Paleogene (Kennett 1982; Lazarus and Caulet 1994). The evolutionary effects of temperature and circulation change may be more apparent in faunas from these polar regions than in the lower latitude planktonic foraminifera studied so far. Thus, it is important to examine patterns of macroevolutionary change in other groups of organisms, particularly in regions outside the warm surface waters of the tropics to subtropics, to gain a more complete understanding of the relationship between environmental change and evolution. Understanding evolutionary processes in polar regions, particularly the Antarctic, is also of interest in itself, as these regions have played major roles in the evolution and maintenance of global biodiversity and global geographic biodiversity gradients, at least in Cenozoic times (Crame 1989; Crame and Clarke 1997).

Radiolarians are one group that can provide such data, as they are relatively diverse (globally, ca. 400 living morphospecies - Casey et al. 1979; Takahashi 1991) and have diverse polar faunas. Radiolarians also inhabit a broader range of water depths and occupy a broader range of ecological niches than do planktonic foraminifera (Casey et al. 1979; Casey 1993; Anderson 1980, 1993). Despite this potential, macroevolutionary studies of radiolarians are rare, and, to the author's knowledge, only two such studies have been published, both of low latitude Mesozoic faunas, that have directly examined the relationship between evolutionary change in radiolarians and environmental change. Erbacher et al. (1996) detected a close correlation between faunal turnover in late Cretaceous Tethyan radiolarian faunas and oceanic anoxic events. Danelian and Johnson (2001) compared patterns of faunal change in Tethyan Jurassic and early Cretaceous radiolarians to the general history of regional paleoceanographic change, and argued for an inverse relationship between oceanic productivity and faunal diversity.

Antarctic Neogene radiolarian faunas are employed in this investigation to explore these issues. These faunas are diverse and have been exposed to major changes in the environment due to long-term trends of cooling and glaciation of the Antarctic continent (Kennett 1982). The Neogene Antarctic radiolarian record is thought to closely reflect the original living diversity of the plankton. As is true for other regions of the ocean with good biosilica preservation, the great majority of living taxa in the Antarctic Ocean are also found in surface sediments, at least in regions not heavily influenced by glacial marine sedimentation (Lozano and Hays 1976; Nakaseko and Nishimura 1982; Abelmann 1992b). Furthermore, except for the earliest Miocene, Neogene dissolution has not been extensive enough to strongly affect preserved radiolarian diversity (Chen 1975; Lazarus 1990, 1992; Abelmann 1990, 1992a; Caulet 1991). Many Antarctic Neogene species are also endemic to the region, and, as is true for most Antarctic plankton, show little within-region geographic restriction (being instead distributed throughout the Southern Ocean, see Hays 1965; Lazarus and Caulet 1994). This is due to the circumpolar circulation system which, as in other plankton provinces, mixes biotas throughout the Antarctic water mass over decadal periods (Kennett 1979; McGowan 1986).

The fossil record of these faunas thus represents a relatively easily sampled, unusually complete history of the evolution of a distinct biota. Due to the availability of detailed records of regional environmental change, this fauna also provides useful test material for examining the role of environmental change in the evolution of faunas. Although our knowledge of these faunas' taxonomy and stratigraphic distribution are still incomplete, extensive study and recovery of new materials over the last decade by the Ocean Drilling Program (ODP) has provided a wealth of new material. This report presents the first quantitative, fauna-based macroevolutionary analysis of the Antarctic Neogene radiolarian record, or for that matter (so far as is known to this author), of any Cenozoic radiolarian fauna. This study is based on the range-chart data of three authors from ODP Legs 119 and 120 to the Kerguelen Plateau: Caulet (1991), Abelmann (1992a), and Lazarus (1992). The primary goal of this work is to document and interpret the correlations, if any, between known patterns of environmental change and evolution of the fauna.

A second, methodological goal is present as well. Many previous studies of macroevolution in the plankton cited above have relied on large compilations of literature data, and frequently on indirect representations of the primary observational data via major taxonomic syntheses (e.g., Kennett and Srinivasan 1983; Pearson 1993) as the base for analyses. Others (e.g., Corfield and Shackleton 1988) have carried out comprehensive surveys of faunas from selected sections as a base for analysis. Neither of these methods can yet be employed in radiolarian work. Major taxonomic/stratigraphic syntheses (at least below the family level) comparable to those for planktonic foraminifera are not yet available for radiolarians. Nor, given the highly incomplete species-level knowledge of fossil radiolarian diversity, is it possible to easily collect comprehensive faunal data from stratigraphic sections. These limitations must be overcome in order to use the data currently available: the original published marine micropaleontological biostratigraphic literature. Direct use of such primary range-chart data has been comparatively limited in macroevolutionary research, despite its richness (the Neptune Database of selected Deep Sea Drilling Project (DSDP)/ODP range chart data for example currently has nearly 400,000 primary observational records in it for several thousand species of Cenozoic plankton (Lazarus 1994). Given both the potential due to the vast extent of this primary data set and the difficulties that can occur in compiling such data, the secondary goal of this study is to explore methods that are appropriate for synthesizing this largely underutilized record of evolutionary change.

Prior Research

Current knowledge of radiolarian faunas in Antarctic Neogene sediments is derived from early taxonomic/stratigraphic studies (Hays 1965; Petrushevskaya 1967, 1975; Chen 1975) of piston and DSDP cores, and more extensive taxonomic, stratigraphic and distributional study from numerous ODP cores, which also generally have accurate geochronology based on both microfossils and paleomagnetic data (Gersonde et al. 1990; Barron et al. 1991; Harwood et al. 1992). However, this research has been done using the constraints imposed by the nature of Southern Ocean deep-sea drilling. The most important constraint has been that, when the drilling ship has been in the Antarctic, several legs have been run in quick succession, resulting in several researchers working on essentially the same radiolarian faunas simultaneously. In early DSDP work this frequently led to multiple, conflicting descriptions of the same species (e.g., Chen 1975; Petrushevskaya 1975, see also discussion in Lazarus 1990). Studies on the more recent ODP materials tended, by contrast, to be divided into individual studies of different, only slightly overlapping geologic time intervals. This reduced the degree of potential taxonomic confusion, given the largely non-overlapping ranges of time being examined, and also reduced the taxonomic scope of individual studies to more manageable levels. Essentially, each researcher compiled a checklist of taxa of interest for his or her geochronologic study interval. Advantageous as this has been for the taxonomic and stratigraphic goals of the original research, the largely non-overlapping time intervals of the published primary data creates a significant complication for later evolutionary analysis.

Despite the much improved knowledge of these faunas as the result of recent work, faunal lists are still very incomplete, with perhaps as many as half the preserved taxa not yet described or studied stratigraphically. This, in itself, is not necessarily a major limitation for study of faunal evolutionary patterns—most studies of macroevolutionary pattern, at least those based on multicellular taxa, are based on much less complete data than used here. What is important however, is that the taxa studied to date are a fair representation of the original total diversity. It is not believed that any major systematic bias exists in the choice of taxa studied so far, although the primary stratigraphic interest of research on these faunas may have somewhat biased the data collection in favor of short ranging taxa. Significantly, at least some taxa from each family-level group appear to have been included in published studies, with roughly similar numbers of taxa included from the two basic divisions of fossil radiolarians (Spumellaria and Nassellaria) as well. However, in the existing data, not all taxa have been studied in all stratigraphic intervals by all authors. In particular, each researcher selected a subset of the taxa available for his or her study, and these nonstandardized checklists are difficult to combine without creating a major compilation artifact: a species studied by one worker may not have been examined by another, despite its presence in the latter's material. This has the data compilation effect of artificially truncating the range of the taxon at the temporal boundary of the study intervals between the two workers. A second problem is the non-standardization of species concepts. Researchers cooperated in creating standardized definitions for known or suspected stratigraphic indicator taxa, but for other taxa coordination was less rigorous, and the scope of taxon names (some still in open nomenclature) can differ significantly from one worker to the next. This can lead to substantially different reported stratigraphic occurrences for taxa, since many of the taxa studied are parts of anagenetically evolving phyletic lineages, and different morphologic boundaries between taxa divide the lineage into named taxa with different temporal ranges.