The observed decline of diversity near the beginning and end of each author's study interval is, of course, to be expected, given the way the data were collected. As the beginning and the end of the study interval was reached, fewer taxa characteristic of the core of the study interval are found. An increasing proportion of the taxa present do not belong to the list of taxa used by the author and are thus not recorded. When interpreting data of this sort it is therefore necessary to distinguish the bias towards lower diversity of this 'edge' effect and place most reliance on the data from within the main study interval of each author. However, in those intervals where data of different authors overlap, shorter-term trends may still be of interest. In particular, they can be compared between data sets to check the consistency of patterns as seen by different authors, and thus provide an indication as to how accurately the data reflect a primary biologic signal instead of only artifacts in recording of data. The intervals in which the data of different authors overlap (early Pliocene, late early/early mid-Miocene, Fig. 3) show a high degree of similarity to each other.

Data from different authors also show a high degree of coherence in most other estimates of evolutionary change. Although there are some differences in absolute values, first and last occurrence patterns and total turnover (Fig. 4), as well as average longevity values for taxa (Table 5) from different time intervals, all show similar results. Most remarkably, even the bimodal distribution of taxon longevities is seen in both Lazarus' and Abelmann's data sets (Fig. 5).

These similarities strongly suggest that the patterns shown in Figure 3, Figure 4, and Figure 5 are not just artifacts of different authors' observational methods, but instead reflect primary patterns of the actual fossil record. Given the well-preserved nature of the faunas and the lack of obvious bias in the choice of taxa studied, it seems reasonable to believe that the patterns observed also reflect original patterns of faunal change in the living faunas. It is therefore valid to compare the patterns of faunal change to those of environmental change to answer the question posed at the beginning of this article: are faunal changes in these planktonic organisms strongly correlated to those of environmental change?

Figure 6 summarizes Kerguelen Plateau Neogene environmental data. Although there are gaps in these records due both to hiatuses in the sections and occasional lack of sufficient foraminifera for isotopic analysis (particularly in the late Neogene), clear patterns are nonetheless apparent. Prior to 15 Ma, oxygen isotopes were fairly stable and percent sedimentary carbonate values were uniform and high. Carbon isotope values were increasing, and planktonic values tended to be heavier than benthic ones. The period between 15 and 13 Ma was one of major change, with short-term, large-amplitude fluctuations appearing in both isotope and carbonate records. Oxygen isotopes shifted towards heavier values, carbon isotopes became more negative, and the offset between benthic and planktonic values decreased. This new regime persisted to approximately 6.5 Ma, at which point there was a dramatic drop in carbonate values. There was also renewed change in both oxygen and carbon isotopes, with both indicators shifting again in the same direction as in the prior shift at 15-13 Ma. These changes reflect the well-known Neogene history of the Southern Ocean, which is dominated by two major events, each of which shifted the system into a new state (Kennett 1982). The changes between 15 and 13 Ma reflect the mid-Miocene increase of glaciation on the Antarctic continent, together with cooling of both surface and deep waters in the Southern Ocean, and an increase in biologic productivity (Flower and Kennett 1993, 1994). The ca. 6.5-4.5 Ma shift reflects further increases in Southern Ocean productivity, a shift towards diatom-dominated primary production, and is associated with a (poorly understood) interval of major glaciation of the Antarctic continent near the Miocene -Pliocene boundary (Van Couvering et al. 1976; Kennett 1982).

Selected evolutionary metrics and environmental records are compared to each other in Figure 7. A clear correlation can be seen between the evolutionary (Fig. 7A-C) and environmental (Fig. 7D-F) data. The overall trends in diversity (Fig. 7C) are correlated to the carbon isotope records (Fig. 7D), particularly that of the planktonic foraminifera, while there is a good general match between levels of turnover (primarily via stepwise increases in last occurrences, Fig. 7B) and the three major phases of Southern Ocean development that are delineated by the two major intervals of rapid environmental change at 15-13 Ma and near the Miocene-Pliocene boundary. Faunal diversity thus appears to be inversely correlated to productivity, as indicated by stable carbon isotope values, and by the increase in biogenic silica in sediments (indicated by the decrease in carbonate content in Fig. 7E). This correspondence is reinforced when other estimates of regional productivity are considered. Sedimentation rates in Antarctic Neogene pelagic sections for example tend to reflect water-column productivity, particularly in those later Neogene biosiliceous sections where carbonate is nearly absent, and thus changes in carbonate dissolution are not important. Although no comprehensive synthesis of regional data is available, sedimentation rates in the Antarctic Neogene have tended to increase from fairly low values in the early Miocene, to moderately high values in many sections in the late Miocene, reaching a peak in the early Pliocene before declining into the later Pliocene and Pleistocene (Brewster 1980; Gersonde et al. 1990; Froelich et al. 1991; Barron et al. 1991; Harwood et al. 1992). This pattern, in inverse form, is also reflected in the biodiversity curve (Fig. 7C).

There is no evidence for a pattern of near uniform, continuous change as predicted by models of evolution driven primarily by biotic interactions (Stenseth and Maynard-Smith 1984; Hoffman and Kitchell 1984). Rates of evolutionary change vary dramatically and are clearly linked temporally to changes in the environment. Environmental change, directly or indirectly, thus appears to play a major role in the evolution of Southern Ocean Neogene radiolarian faunas, with the primary timing and general magnitude of evolutionary change being driven by external environmental forces. These results may also be relevant to understanding factors driving the evolution of Antarctic biotas in general. As noted by Clarke (1990), temperature change and cold absolute values for temperature, do not by themselves seem to be sufficient explanations for extinctions of taxa in the historical development of Antarctic faunas. It may be instead that for much of the Antarctic biota, and not just for radiolarians, extinction has been caused by changes—primarily increases—in oceanic productivity, which according to Hallock's (1987) model, decreased the range of habitats in the 'trophic resource continuum', and thus led to the loss of biodiversity.

One of the more interesting, and quite unexpected, results of this study are the patterns of longevity data for species. All data show a similar pattern of bimodal longevities, separated by the relative dearth of taxa with longevities in the range 10-12 m.y. The subpopulation of long-lived taxa seems to be found primarily in species that originate in the older part of the study interval, below 13 Ma, while species of more recent origin are notably shorter lived. One possible explanation for such a bimodal pattern would be the existence of two distinct evolutionary sub-groups within the radiolarian plankton, such as might hypothetically exist between surface and deep-dwelling faunas. Radiolarians in many parts of the modern ocean include both shallower dwelling forms and deeper-dwelling species (Casey et al. 1979; Takahashi 1991; Casey 1993). The Southern Ocean, however, is different from lower latitude environments in that modern radiolarian faunas are most abundant in deeper waters (>150 m to 1,000 m or more) (Abelmann and Gowing 1996). In the early Neogene, prior to the mid-Miocene climate shift, surface waters in the Antarctic may have more closely resembled those of modern cool-temperate regions, with relatively more radiolarian taxa living in the uppermost layers of the water column. The observed bimodality in evolutionary rates in the older Neogene data would thus reflect differences in characteristic evolutionary rates between these two depth environments. Depth-related differences in evolutionary rates are also known from lower latitude planktonic foraminifera (Stanley et al. 1988), where deeper dwelling forms were found to have shorter mean longevities than the shallower taxa, possibly due to the lower abundances (population sizes) of deeper dwelling taxa. Extending this idea to the current study would suggest that the longer lived mode seen in the older Neogene radiolarian longevity data reflect near-surface dwelling forms. With the loss of this near-surface water environment in the mid-Miocene, later Neogene radiolarian faunas became primarily deeper dwellers. The longevity data reflect this shift in that only shorter lived forms are seen in the late Neogene longevity histograms. Some support for this interpretation also comes from the changes in taxonomic abundance between Nassellaria and Spumellaria (see below).

The observed pattern, however, can also be explained in at least two other ways. First, the bimodality must—at least partly—reflect the time-windowed nature of these data, plus the substantial extinction that occurred between 15 and 13 Ma. Many of the taxa lost at the mid-Miocene event exhibit truncated ranges at the base of the study and thus are constrained to have longevities <10 m.y. The distribution of the survivors' longevities suggests that many that do survive then continue to near the second, Miocene-Pliocene boundary, extinction event, before becoming extinct, thus creating a second mode of long-lived taxa. However, the same type of explanation cannot be used to describe the longevity pattern of taxa originating after 13 Ma, as they show a significantly different distribution of longevities, with no evidence of a large number of long-lived taxa that go through the end-Miocene event and continue into the Recent. The post mid-Miocene fauna therefore consists of (on average) significantly shorter ranging species, implying an increase in evolutionary rates in the region during the Neogene interval studied.

A second possible alternate explanation is that not all Southern Ocean taxa are in fact endemic, and in particular, significant numbers of the post-Miocene taxa appear to be bipolar, with records of occurrence as well in the Norwegian-Greenland Seas (Bjørklund 1976). Some of the turnover, particularly in the late Neogene, may thus represent immigration of taxa. This is still of interest but represents a different biological process for diversity regulation than in situ evolution (see also discussion in Hoffman and Kitchell 1984).

It should also be noted that both the overall rates of change (up to more than 50% species turnover/m.y.) and the average longevity of species (4.8 m.y.) are comparable to, or exceed, rates that have been previously reported for marine microplankton groups (ca. 7-16 m.y. for cladogenetically delimited lineages of planktonic foraminifera (Stanley et al. 1988); 5.6 and 9.5 m.y. mean longevity for Cretaceous to Recent keeled and non-keeled species of planktonic foraminifera (including both cladogenetic and anagenetic evolution, Norris 1991), and ca. 5-10 m.y. for species of planktonic foraminifera, coccolithophores and radiolarians, based on the survival decay curves given in Hoffman and Kitchell (1984).

Although external environmental change appears to be the primary factor driving evolution of Antarctic Neogene radiolarian faunas, biology does seem to play an important role. This is suggested by two aspects of the results. First, diversity and evolutionary turnover metrics appear to correlate more closely to productivity (carbon isotopes, sedimentation rates), and the composition of the primary producer flora (carbonate/biogenic silica ratio), than to temperature itself (oxygen isotopes). This result is similar to those from the study of the warm water Cenozoic planktonic foraminifera, (e. g., Corfield and Shackleton 1988; Lipps 1986; see also discussion in Norris 1991) and Tethyan Mesozoic radiolaria (Erbacher et al. 1996; Danelian and Johnson 2001), and lends general support to Hallock's (1987) trophic model of biodiversity regulation. Second, the shift from a lower productivity, carbonate phytoplankton ocean to a higher productivity, biogenic silica phytoplankton ocean is accompanied by a shift in the ecological nature of the radiolarian fauna (Table 5). Spumellarians are more common in the older Neogene but are less common in the later Neogene. Spumellarians are, as a group, more likely to host symbiotic algae and to live in near-surface waters than are Nassellaria, which tend to be deeper dwelling, non-symbiotic forms (Casey et al. 1979; Takahashi 1991; Anderson 1980, 1993; Abelmann 1992b). The shift from spumellarian dominated to nassellarian dominated faunas suggests a shift from faunas more adapted to lower productivity, silica-rich surface waters, to faunas living in deeper waters, below a relatively silica poor surface water layer dominated by highly productive, silica depleting diatoms. Diatoms are known to be particularly efficient in removing available biogenic silica in surface waters, and thus are thought to have a potential negative effect on radiolarian growth (Anderson 1980; Takahashi 1991). On a much broader scale, Harper and Knoll (1975) suggested that the Cenozoic evolution of both diatoms and radiolaria has been influenced by competition for silica. However, it should be noted that in the modern Southern Ocean, biogenic silica is not a limiting nutrient - light and water column stability play instead more important roles (Siegfried et al. 1985). Whether this condition was also true for earlier intervals in the Neogene is not known.

Antarctic Neogene radiolarian faunal occurrence data drawn from several recent ODP studies have been used to reconstruct the diversity and evolutionary turnover history of this group, and to compare the patterns of environmental change. A clear correlation exists between overall radiolarian diversity and proxies for ocean productivity, such as stable carbon isotopes, regional sedimentation rates, and abundant biogenic silica in sediments. Higher rates of inferred productivity characterize the later Neogene and are associated with lower radiolarian diversity, and an increased dominance of Nassellarians, a predominantly deeper dwelling group of radiolarians with fewer photosymbiont taxa.

Overall rates of evolutionary turnover, and particularly rates of extinction, are not uniform, but are also closely linked to environmental change, with shifts towards increased turnover and extinction rates being associated with the shifts in the environment marked by the mid-Miocene and end Miocene glacial events that affected the Antarctic region.

The average longevity of species also has been affected by environmental change, with post mid-Miocene species having significantly shorter stratigraphical ranges than older taxa. This seeming acceleration of evolution is a phenomenon that has not been reported, or at least much emphasized in previous studies of warmer water microplankton. Norris (1991), for example, presents data suggesting that longevity distributions in Cretaceous to Recent planktonic foraminifera have remained largely unchanged for the last ca. 100 m.y.

These results together suggest that changes in the environment, not internal biologic interactions, primarily regulate overall rates of faunal change in Antarctic Neogene radiolarians. This largely confirms prior results from study of lower latitude faunas of planktonic foraminifera and radiolaria, although changing average longevity distributions in Antarctic Neogene radiolarian faunas suggest that important differences may exist as well, and that the environmental history of polar and low-latitude regions may differ in evolutionarily significant ways. Lastly, biology, although not the ultimate pacesetter of evolutionary change, nonetheless plays an important role in mediating changes in the physical environment through mixed physical-biologic mechanisms such as change in marine productivity.

This study is a preliminary analysis, as the available taxonomic, stratigraphic and biologic data on these faunas are still very incomplete. Clearly, more complete surveys of the faunas using existing taxonomy, as well as a more complete taxonomy, would be desirable. So would a better knowledge of the ecology of the taxa studied, particularly the relationship between different taxa and marine productivity, including the distribution of symbionts among species. The relative importance of in situ evolution vs. immigration from other cold water regions also needs to be resolved. However, despite these limitations, it appears that useful evolutionary signals can be extracted even from relatively incomplete data such as that recorded in normal DSDP/ODP radiolarian biostratigraphic range charts. Given that few studies of evolution have been carried out with such data sets, despite their abundance in the published literature, it is hoped that additional studies may be possible which will yield further insights into patterns and processes of evolution in the pelagic realm.