4. RESEARCH PROJECTS CARRIED OUT WITH NEPTUNE, WHAT THEY TOLD US, AND RECOMMENDATIONS FOR THE FUTURE

4.1. The database structure and search capabilities: a tool to find out what we do or don’t know

The Neptune database currently provides rapid retrieval of age information on 165 DSDP and ODP holes; taxonomically corrected species lists and other taxonomic information for calcareous nannofossils, planktic foraminifera, diatoms and radiolarians for the entire Cenozoic; paleogeographic location of the 165 holes (paleolatitude and paleolongitude); extensive distributional data for these fossil groups (e.g., biogeographic occurrence information, computerized microfossil range charts) (Fig. 4.1). The design and implementation of the database software have been described in Lazarus (1994) and in an unpublished guide (Lazarus, personal commun., 1996). These will be used for this description, with the updates of the data tables based on the present status of the database (after the most recent upgrade).

4.1.1. Overview of database structure. Neptune is designed as a relational database. Macintosh computers and 4th Dimension™ database software (4D) are used to run the database (Lazarus 1994). The database is implemented as several relational tables that contain (as of February 1998) close to 500,000 records.

Import procedures for range chart data as well as search procedures are available. The search procedures can locate all reported occurrences of any taxon or combination of taxa, automatically identifying occurrences recorded under synonymous names. Searches can also be used to locate other relevant information, such as general hole information, sample age, species occurrences, etc. Commercial mapping software (e.g., Atlas™) is used to plot locations of species occurrences, using a Neptune-generated plotting data file with latitude and longitude. A ‘composite age range chart’ program can also be used with an appropriately formatted file generated by a Neptune search (Lazarus 1994).

Neptune was created as a relational database where the data are separated into simple tables, with relational links between the tables. The structure of the database is shown in Figure 4.2. Five data tables hold the primary data: stratigraphic occurrence data for taxa (‘Bug Data’ table); taxonomic data on species’ names (‘Taxonomy’); biogeographic data on species’ occurrences (‘Taxa by Hole’); geologic age information (‘Age Models’); and paleogeographic information (‘Geographic Info’). Paleo-water depths are available for selected samples and have been published (Spencer-Cervato 1998). As this information is not available for all samples, it is not currently included in Neptune.

The database maintains a strict separation between the primary observational data (occurrences of named taxa at specific depths in holes) and the interpreted meaning of the data (i.e., the species to which the name belongs - ‘Taxonomy’ - or the age of the section at a specific depth - ‘Age Models’). These tables can be in fact modified repeatedly, but the observations remain constant (Lazarus 1994).

The stratigraphic occurrence data form the core of the database (over 380,000 records of ‘Bug Data’). A typical range chart is decomposed into a minimum of one species occurrence in one sample. Further data separation is achieved by putting all information about samples and species into separate tables (‘Sample Data’ and ‘Taxonomy’). Samples and species are represented in the ‘Bug Data’ table only by internal codes, linked to the more detailed records in other tables.

4.1.1.1. Species names (‘Taxonomy’ table). All names in the database are identified by a separate entry in this table. Each occurrence, including misspellings and questionable names (e.g., A. deflandrei?), is given as a separate entry. Each is identified by up to three words (genus, species, subspecies or qualifier). The qualifier is generally used to identify questionable entries, marked with the letter ‘Q’. Each entry is uniquely identified by a ‘Taxon code’, a combination of nine characters originally given by DSDP. This code is central to the functioning of the database, as it provides links to the other tables. The first five characters are letters, all upper case. The first letter identifies the fossil group (D for Diatoms, N for Nannofossils, etc.). The next four letters are characteristic of the genus. The last four characters are numbers and give the species number in the genus. DSDP started with 0010 and incremented by units of 10 for each new species name. ODP does not use codes to identify species names. Thus, we created new codes for new species names that occur in ODP range charts. To avoid any overlap, we have used the same 5 letters to identify the group and the genus, but started with 5010 to number the new species (DSDP never had numbers higher than 2500).

Every entry has other information attached to it. The Status (or validity) field is a single uppercase letter which states that the name is V-alid, a S-ynonym to another name (with corresponding taxon code entered in the ‘synonymous to’ field), I-nvalid, Q-uestionable, or U-nknown. A G is used to indicate a genus-level name. Every name has also an author code (initials of person responsible for the entries in the Status field) and a date (mm/dd/yy). Comments of any length are also entered in the ‘Comments’ field.

Additional species’ records are available in the ‘Species by Hole’ table which comes directly from the DSDP data set. The current table, reformatted from the original data set to save space, contains simply a Taxon Code and a Hole field.

4.1.1.2. Age Models and Hole summary data (‘Age Models’ and ‘Hole Data’ tables). The ‘Age Models’ table contains the age model developed for each hole with range chart available in the database. The age model was constructed by a broken line composed of straight segments, which can be horizontal in the case of hiati. The extremes of the segments are identified by age and depth and entered in the corresponding field, next to the ‘Hole’ field. To keep track of which age model is being used in the database, each age model’s time of creation date stamp is entered automatically in the ‘Hole Data’ table (‘Age model version’). This latter table contains a variety of information, including latitude and longitude, water depth, ocean basin, hole length and recovery, etc. Holes that have an age model (and therefore range chart data) have an entry in the ‘Age Model version’ field and a ranking for each fossil group (originally used to select holes).

4.1.1.3. Sample Info (‘Sample Data’ table). Most of the fields in this table are created directly from computer files or by Neptune. Each sample described in each range chart is identified by a unique digital code and is specific for one fossil group. This means that, if in one sample (identified as depth in a hole (mbsf), but also in three separate fields as core-section-depth interval format - grouped in Figure 4.2 under ‘Sample Name’) both diatoms and radiolarians (‘Taxonomic Group’ field) were described, this sample would be described twice in Neptune, each time with a different ‘Sample Code’. The age of the sample is derived from the ‘Age Model’ table through a relational link. If available, information on the preservation and abundance of the specific fossil group in that sample is also given.

4.1.1.4. Paleogeographic data (‘Geographic Info’ table). With the addition of Paleogene range charts, I considered it necessary to locate species occurrences in their appropriate paleogeographic position. For this purpose, I used a PC-based program kindly provided by Alan Smith (Cambridge University) which uses finite rotations. The program is based on published reconstruction data (Euler rotations and their ages) used to move a given site relative to Africa and then reposition that site in paleomagnetic coordinates (Smith, personal commun., 1997). The input file contained present latitude and longitude: paleolatitudes and paleolongitudes were determined for each hole at 5 m.y. intervals. This approximation was necessary to simplify the entry of these data into Neptune, but I believe that it does not significantly affect the already approximated estimate of paleolatitude and paleolongitude made by the finite rotations program. These paleogeographic data, with hole and age, were imported into a separate table (‘Geographic Info’) and the Sample Code used to link it to other tables.

4.1.2. Importing data into Neptune (range charts and age models). Data can be imported into Neptune by the ‘administrator’ (this function is not available in the ‘user’ mode or with the runtime version of 4D). Most of the DSDP range charts were imported directly from the DSDP CD-ROM data, but ODP data need to be imported as individual spreadsheet format files. Each procedure creates automatically new sample records for each sample in the range chart data (‘Sample Data’) and new ‘Bug Data’ records for each non blank cell of occurrence data in the range chart. No ages are assigned in this procedure and all sample ages are set at zero. Only when the age model is imported, a corresponding age is recorded in the age field of the ‘Sample Data’.

Each range chart file needs to hold data for one hole and one fossil group only. Each sample must be entered in one row in a ‘leg-hole-core-section-first depth-second depth’ format (e.g., 113-689-B-2H-1-115-116). These data are automatically entered in the corresponding fields in the ‘Sample Data’ table. The depth in mbsf is derived from the ‘Core Data’ table, where the core depth files for each hole are imported as soon as a hole is selected. Species names must be entered as Species Codes (9 characters, e.g., DACTI0020). Every species code present in the spreadsheet must be already available in the ‘Taxonomy’ table. The ‘import from spreadsheet’ procedure in Neptune automatically checks each DSDP Code in the spreadsheet and if it encounters a code that is not present in the ‘Taxonomy’ table, the procedure aborts.

Age information is present in Neptune in two forms. The ‘Age Model’ table actually holds all the line of correlations (age models) for each hole. Age for samples are calculated from the line of correlations and stored as calculated fields in the ‘Sample Data’ table. This calculation is done only once, when the age model is read into the database, and is automatic. Only one age model can be imported at a time. To update an existing age model, it is sufficient to read in the new file and the old ages will be automatically replaced by ages based on the new line of correlation.

4.1.3. Report capabilities and external graphics. Data can be extracted from the database in a variety of ways. The results can be then saved as export files, that can eventually be used with other programs. Procedures that search for taxa, in either the ‘Bug Data’ table of stratigraphic occurrence information, or in the ‘Taxa by Hole’ table of biogeographic information, create lists of Taxon Codes (from ‘Taxonomy’) to search for. These lists include the taxonomic name/s requested by the user, but are supplemented by lists of synonyms to these names. Users can edit these lists to fine-tune searches.

In addition to export formats for statistics and spreadsheet packages (usually in ASCII format), the database exports data in formats specific for two types of graphic data display. Data on the location of specific DSDP/ODP sites can be plotted in a map form using Atlas™ (WTC Scientific). The most recent version of this program for Macintosh computers does not run reliably on PowerPCs and the use of the PC IBM-compatible version (which can use the same cross-platform file) is recommended. A custom application creates graphic displays of occurrence data for taxa, plotted by age and hole (‘Age Range Charts’, Lazarus 1994) (Fig. 4.3).

4.1.4. Searching the database. The simplest way to search Neptune is by using the built-in 4D ‘Search Editor’ (under the ‘Select’ menu). Any of the tables previously described (Species Names, Hole Info etc.) can be selected from the list in the small window that automatically appears when Neptune is started. This shows a window with all the records in that table. The ‘Search Editor’ function displays a dialogue window which shows the fields available in the table. Only fields in bold can be selected and additional search criteria (equal to, contains, less than etc.) added. The results of the search are displayed in a few seconds (Fig. 4.4).

More complex procedures, such as a ‘Bug Data’ search, allow to locate range chart data about one or more taxa. These predetermined procedures can be selected with the ‘Execute Procedure’ function under the ‘Special’ menu (Fig. 4.5). The ‘Bug Data Search’ procedure first shows a search editor window for ‘Species Names’ and waits for a taxon entry. This can be formulated as ‘Species - is equal to - name’ or done directly with DSDP codes (Fig. 4.6). This procedure locates all taxa matching the entered criteria, as well as other taxa identified in the database as synonyms for any of these. The user can then select one or all of the identified taxon names and click the ‘done’ button at the bottom of the window (Fig. 4.7). The procedure then searches the ‘Bug Data’ table to locate all records for this list of taxa. This search is done using indices, and only takes a few seconds (Fig. 4.8). The procedure informs the user via a dialogue box how many records have been found, and then presents the search editor window a second time. At this point the user can enter any other criteria, such as only samples with ages greater than 0 (i.e., holes with age models), or from holes from a specific geographic location (Fig. 4.9). The 'search in selection' box (lower left corner) is automatically marked allowing to search only among the already identified occurrences (and not the whole database!). The user should then click 'ok' to proceed. The procedure will refine the selection according to these secondary criteria, and present the user with a list (Fig. 4.10). This list can also be edited to refine the selection. Lastly, the user clicks 'done' to exit the procedure. The selected records can then be printed, exported to disc, or summarized in a report.

A search procedure is also available to the automatic search for the ages of all samples recording several taxa given in a list (and their synonyms). This ‘Batch Search’ (‘BugDataSearchBat’) allows for the automatic operation of the series of procedures described above (Species name selection, identification of synonyms, bug data search, restriction to holes with age models, sorting of samples by age). It produces one separate output file for each name, as well as a cumulative file. This procedure was used to obtain species longevity data (described below). Alternatively, samples can be sorted by latitude and longitude to obtain ranges of geographic distribution of taxa through time (e.g., to identify cosmopolitan or endemic taxa).

4.2. Paleontological research based on Neptune: plankton evolution

In this and in the next section (4.3), I am presenting a summary of published paleontological and stratigraphic research conducted with Neptune, as well as some unpublished data on macroevolution. Neptune’s potential for paleontological research has been, so far, only marginally exploited. In spite of the limitations outlined in Chapter 3, the database provides the opportunity for large-scale macroevolutionary studies that could go well beyond presently available studies (e.g., Jablonski 1993; Kammer et al. 1997). The age control and time resolution, combined with the taxonomic information at species level on four distinct plankton groups, make Neptune a high quality data set.

Currently, the two studies we published on evolution were focused on the evolution of one foraminifer species and were based on Plio-Pleistocene sediments, which are represented in a large number of holes in Neptune (Lazarus et al. 1995b; Spencer-Cervato and Thierstein 1997). The goal of these studies was to document patterns of evolution of a new species (the planktic foraminifer, Globorotalia truncatulinoides) from its ancestors and to identify speciation and migration in distinct biogeographic provinces by using the tests’ morphometry. In addition, we attempted to determine the environmental conditions (water depth, thermal structure of water column) at the time of speciation or immigration with stable isotope geochemistry. Whether changes in these environmental conditions were a determinant factor in the speciation or migration, even after these detailed studies, remains still speculative.

For these studies, Neptune was used in the selection of sites by identifying the occurrences by hole and FADs/LADs (first appearance datum/last appearance datum) of the species and its ancestors. An age range chart was produced from the search for all G. truncatulinoides and related species occurrences (Fig. 4.3). This was used to identify the oldest first occurrences and to have an overview of the age distribution, which shows a distinct diachrony (Fig. 4.11). This search lead to the selection of suitable DSDP and ODP sites from which samples were requested. The samples were then analyzed morphometrically and isotopically to determine patterns of evolution (in this case, cladogenesis or phylogenetic branching) and species migration (Lazarus et al. 1995b; Spencer-Cervato and Thierstein 1997).

The earliest first occurrences are found in several sites in the southwest Pacific. Gradual cladogenesis was documented in this region during the late Pliocene in sympatric or parapatric populations (Lazarus et al. 1995b). Based on qualitative observations, similar but younger, gradual transitions had been reported from other areas of the world’s oceans. Therefore, the hypothesis arose that this gradual evolutionary branching might have occurred in response to changing environments at different times in different ocean areas. To evaluate this hypothesis, we studied the morphological transitions of the three taxa, using image analytical techniques, in several deep-sea sections from various areas, identified with Neptune (Spencer-Cervato and Thierstein 1997). The morphometric analyses showed that G. truncatulinoides evolved between 2.8-2.3 Ma sympatrically in large populations from its ancestor G. crassaformis in the southwest Pacific. Differentiated morphotypes of G. truncatulinoides subsequently immigrated into the Indian and Atlantic Oceans between 2.3 and 1.9 Ma. Our morphometric data show these younger appearances outside the southwest Pacific to be punctuated, and representing migration events (Spencer-Cervato and Thierstein 1997).

One of the most crucial, yet elusive issues in evolution is the role played by the environment in the appearance of a new species or its extinction. Planktic foraminifera are ideally suited for these studies because of the large populations, widespread occurrence of tests in marine sediments, relatively large size that allows for detailed identification with traditional microscopic techniques, abundance of information on living populations and their habitats, conceivably rapid colonization of biogeographic provinces under suitable conditions, and the possibility to reconstruct these conditions (e.g., water depth, temperature, nutrients) with stable isotope geochemistry.

Globorotalia truncatulinoides is an ideal species for the study of the environmental conditions at the time of speciation. We hypothesized that the global cooling of surface waters, coinciding with the northern hemisphere glaciation, led to the formation of oceanographic barriers that could have retarded the expansion of G. truncatulinoides up to 2.3 Ma. At this time, a relative warming and subsequent transgression could have spurred the migration from the southwest Pacific into the Indian Ocean, possibly through the Indonesian passage. A direct link between the speciation and surface water changes linked to the northern hemisphere glaciation has not been proven so far and seems unlikely. In fact, stable isotope data in G. truncatulinoides and its ancestors indicate that the three species’ depth habitat preferences remained unchanged through the speciation and migration of G. truncatulinoides and that all three species were dominantly deep-dwellers, in agreement with their present environmental preferences (Spencer-Cervato and Thierstein 1997).

One of the original goals of the Neptune project was to perform macroevolutionary studies. Macroevolution is a major area of paleontology that developed during the 1970s and 1980s, inspired by the apparent success of the taxic approach to evolutionary patterns (Smith 1994). Macroevolution covers various concepts and processes. These studies differ from the previously described, ‘microevolutionary’ ones - which concentrated on the heritable variations of a population composed of one species and its immediate ancestors - mainly in the scale. Macroevolution studies large-scale patterns of diversification and extinction arising from processes active at or above the species level (Smith 1994 and citations therein). Some workers have instead defined macroevolution as the extrapolation of microevolutionary processes into geological time (e.g., Levinton 1988). Smith (1994, Chapter 4) presents a comprehensive review of macroevolutionary concepts and theories.

Existing studies mainly consider fossil records of marine invertebrates in high hierarchical groupings (orders, families) with low stratigraphic age resolution (e.g., Jablonski 1993) from punctual, geographically restricted sources. Amongst the various causes of artifacts in macroevolutionary patterns, sampling resolution seems to be an important biasing factor (e.g., Alroy 1998). The chronological control of Neptune and its large amount of paleontological data with taxonomic accuracy at species level, combined with its comparably high sampling resolution (on average, 185 k.y. for the Neogene (Spencer-Cervato et al. 1994) and 330 k.y. for the Paleogene) holds promise for potentially significant contributions to this field of paleontology. Because the quality of the results of macroevolutionary studies is strongly dependent on a sound basis of chronology and taxonomy, we have first exploited the stratigraphic data set (see Section 4.3) and revised the taxonomy of the paleontological records (Section 3.3). We have thus left the study of macroevolutionary patterns in marine plankton to the final phase of the project.

I am presenting here some examples of data searches conducted with Neptune to answer some typical paleobiological questions. They cover the overall longevity and speciation/extinction distribution of Cenozoic marine plankton. I am purposely leaving the discussion and interpretation of these data to an absolute minimum. My goal is in fact to show what type of data can be obtained from the database and the potential of Neptune for paleobiological research.

To optimize the diverse paleontological data set in Neptune, the analyses have been conducted separately for the four plankton groups and the results interpreted in terms of similarities or differences among the groups. The data presented here are based on more than 1400 valid species names (Table 4.1), and include the occurrences of their synonyms. The output of every species’ search consisted of their oldest first appearance and their youngest last appearance. Every result was checked to eliminate false entries caused by, for example, occurrences near hiati, typos, occurrences reported in one single sample, etc. Species that were reported only in one hole were not considered to eliminate the bias of single geographic data points. Finally, one table was produced for each group including the species name, the number of times it had been reported in a sample, the location (paleolatitude and paleolongitude) and age of its first appearance, and the location and age of its last occurrence. From these ages, the species’ longevity was calculated.

Figure 4.12 shows the species’ longevity distribution of the four groups with a 1 m.y. resolution. Comparing the four groups, three simple observations can be made: (1) all groups show an asymmetric, unimodal distribution, with a mode around 7 m.y. (diatoms and radiolarians), 14 m.y. (foraminifera) and 19 m.y. (nannofossils), and a tail towards higher longevity values; (2) the median for all distributions is around 10 m.y., except for diatoms, where it is around 7 m.y. - comparing these values with the mode, the peak of the distribution of diatoms is narrower and has a higher symmetry than the other groups; (3) a few phytoplankton species (diatoms and nannofossils) are very longevous (more than 40 m.y.), whilst zooplankton species (foraminifera and radiolarians) live all less than 43 m.y. These observations point to similarities between phyto- or zooplankton in one case, and between siliceous or calcareous plankton in another. However, they also show that diatoms are quite distinct from the other groups.

Table 4.2 shows the average longevity (and standard deviations) of both extant and extinct plankton species. It is noticeable that the longevity of extinct species is consistently shorter than the one of extant species. This could be due to the different sizes of the populations considered (less than 30% of the species are extant), which might also explain the larger standard deviations of extant species’ longevity. Alternatively, this could be the effect of differential preservation. Or it could be caused by the artificial boundary set at the beginning of the Cenozoic - the data might include species originated in the Mesozoic, giving them a shorter-than-real longevity. However, only a very small number of species (e.g., the extant nannofossils Braarudosphera bigelowii and Scapholithus fossilis, the extinct nannofossil Placozygus sigmoides; Perch-Nielsen 1985) are reported also from the Mesozoic. Statistically, they should not significantly affect the data set.

For demographic reasons, one would expect a gradually decreasing longevity instead of the asymmetrical peaks shown in Fig. 4.12. Is this lower-than-expected number of short-lived species an artifact of the analysis or a real signal? On the other hand, the differences seen in the longevity data could be real, suggesting for example that species which evolved in the Neogene (the majority of the extant species) are more likely to live longer. One can only speculate on the cause of this, such as larger surface water temperature gradients linked to growth of ice caps in polar regions? However, the Neogene climate mode, characterized by abrupt shifts from glacials to interglacials and vice versa, would seem to provide stressful environmental conditions that intuitively should increase species turnover, i.e., shorter longevities. This question requires further analyses (e.g., longevity plots at selected critical times) before a viable hypothesis can be formulated.

To help answer the various open hypotheses on species’ longevities, an important factor that should be considered is the geographic distribution of species throughout their duration. This parameter allows to identify endemic versus cosmopolitan species and is an important factor in ecological studies. By comparing this parameter with species longevity, one would test if a species restricted to a narrow geographic region is more likely to survive longer than a globally widespread species, or vice versa.

Appearance and extinction rates were calculated for the four groups to eliminate the bias of the sample size (Wei and Kennett 1983). The rates are calculated as the ratio between the number of extinctions or appearances and the total diversity (number of species) in each 1 m.y. time slice. The Cenozoic appearance rates are shown in Fig. 4.13. Appearances are widespread throughout the Cenozoic and no specific time interval is characterized by anomalously high appearance rates, with the exception of the Paleocene. The graphs show that diatom, radiolarian and foraminifer species appeared all during the Cenozoic, with 100% peaks in the Paleocene, while only 50% of the nannofossil species present in the first million year of the Cenozoic appeared then - the remaining 50% existed already in the Mesozoic (see above for some examples). The apparent late appearance of radiolarians in the early Cenozoic is probably an artifact of the data set: no radiolarian reports are available for the Paleocene (Fig. 3.5). Average appearance rates are less than 10% and only rarely reach 30%, and are characterized by short fluctuations with a somewhat random frequency. In some instances (e.g., at 61 Ma, 35 Ma, 10 Ma), peaks of appearances in one group correspond to peaks in other groups, but no consistent pattern is apparent.

Extinction rate values are much lower than appearance rates and show a more random distribution (Fig. 4.14). Diatoms show a distinct peak in extinctions at the Paleocene/Eocene boundary which is not clearly reproduced in the other groups. On the other hand, radiolarians, nannofossils and foraminifera show a minor peak around the Oligocene/Miocene boundary (25-22 Ma), while all groups (with the exception of radiolarians) show exceptionally high extinctions in the past 3 m.y.

It is interesting to notice that these trends do not correspond to peaks in appearances (Fig. 4.13), but there seems to be a time lapse of a couple of million years between the peaks in extinctions and appearances as the two curves are mostly out of phase. The increase in extinctions in the last 5-7 Ma might be related to the onset of highly variable environmental conditions, which apparently did not cause a corresponding increase in the rate of species’ appearances.

Speciation centers and survival refugia are discrete geographic regions with high concentrations of appearances or extinctions. These are often associated with particularly favorable or stressed environmental conditions and may be limited by biogeographic or oceanographic boundaries (e.g., Jablonski 1993). One simple way to identify these regions is by plotting the latitude of the location of the earliest first appearance or the latest last appearance (Fig. 4.15). The latitudinal distribution of FADs (and of LADs, not shown here, but with an identical pattern to FADs) is clearly different in siliceous and calcareous plankton. Appearances and extinctions of diatom and radiolarian species are concentrated in three belts, around the equator and at mid- to high northern and southern latitudes respectively. These belts are bound by well established nutrient boundaries, like e.g., the polar front. This pattern also reflects the present distribution of siliceous plankton in marine sediments (e.g., Leinen et al. 1986), suggesting that the environmental preferences of these organisms did not change through time. A different scenario is presented by calcareous plankton groups, whose appearances (and extinctions) are more uniformly distributed throughout the latitudinal range.

Even with the limitations summarized in Chapter 3, I attempted to estimate the distribution of plankton species’ diversity during the Cenozoic. This ‘partial’ diversity, limited mainly to occurrences of biostratigraphic markers and biased by the low number of extensive range charts published for DSDP and ODP holes, is still a very comprehensive estimate, even though not a ‘real’ diversity. I present here some preliminary results based on the data included in Neptune.

The total species richness for the four plankton groups was calculated at one million-year intervals. To eliminate the bias of the uneven distribution of the number of sections in Neptune (progressively more sections in younger times, Fig. 3.2), I have normalized the diversity by dividing it by the total number of sections in each time interval. The results are shown in Figure 4.16. The normalized diversity patterns shown by siliceous plankton are quite similar and clearly distinct from what is shown by calcareous plankton. The two distinct patterns shown by the siliceous and calcareous plankton groups are exactly out of phase, with diatoms and radiolarians showing maximum diversity in the Oligocene to Recent, when nannofossils and foraminifera show their minimum values. Both diatoms and radiolarians show a gradual increase in diversity peaking around the Eocene/Oligocene boundary, followed by a relatively stable plateau during the Neogene. Diatoms also show a peak of diversity in the late Paleocene, when radiolarians are not reported. Diversity of nannofossils and foraminifera, instead, peaked during the early to middle Eocene, decreased in the late Eocene, and has remained more of less constant since the Oligocene.

While it is possible that much of the general variability is due to taphonomy, several further speculations could be made on these patterns. However, potential biases would have to be examined first. For example, what is the lithology of the sections in Neptune through time? Are siliceous sediments more common in the Neogene, thereby explaining the higher siliceous plankton diversity? And how do the absolute normalized values compare? The highest values are recorded in nannofossils while the lowest ones are given for diatoms. The number of valid species names in the two groups is almost identical (Table 4.1), but nannofossil names (valid and non valid) are overall slightly more abundant than diatom names in Neptune (Fig. 3.4). However, foraminifera and radiolarian names are the most abundant ones of the four groups, while their normalized diversity is intermediate between diatoms’ and nannofossils’ diversity. Is there a consistent bias in the published range charts, with more reports available on siliceous plankton than on calcareous plankton in the Neogene? The number of reports on Paleogene sections (Fig. 3.5) shows a relatively lower number of reports on siliceous groups than on calcareous groups. The high correlation shown between diatoms’ and radiolarians’ species richness (and to a lesser extent foraminifera) and the total number of sections available for each time interval (Fig. 4.17) suggests that absolute values of species diversity are strongly biased by the size of the data set (i.e., more species are described when more sections, and therefore reports, are available). On the other hand, the species richness of nannofossils shows a complex polynomial correlation with the number of sections but a completely random linear correlation (R2 = 0.07). This may be due to the fact that the correlation is made with the total number of sections and not with the number of sections that contain nannofossil stratigraphy. For nannofossils and foraminifera the latter may be the significant parameter which would perhaps show a higher correlation with the species richness, similarly to the one shown for siliceous plankton.

Finally, how strong is the bias caused by the dominant presence of stratigraphic markers in the reports? Siliceous plankton biostratigraphy is better developed for Neogene sediments than it is for the Paleogene, while it is more uniform for calcareous plankton groups. One approach to this question would be to separate the species included in the distribution into stratigraphic markers, other common taxa and rare taxa, and see if the diversity patterns remain the same or change substantially.

These are only some of the factors that one must consider before a feasible interpretation of these trends can be formulated, and some of these require the addition of data to Neptune which are not currently available (e.g., distribution of siliceous versus calcareous sediments). But however preliminary and partial, these results are still quite encouraging and represent a more detailed data set than what is available from the paleobiological literature.

4.2.1. Availability of relational databases for the paleontological community: the ODP database JANUS versus Neptune. At present, the ODP database, JANUS, which is currently available onboard the JOIDES Resolution and through the WWW, does not represent a viable substitute for Neptune. I must point out, however, that JANUS is very new and that the import of data has just begun. My experience with JANUS is limited to a superficial browsing through ODP’s database WWW site, which provided me with the following information. Site data (water depth, coordinates, length drilled, length recovered, etc.), physical properties (e.g., GRAPE, magnetic susceptibility), and chemical results (e.g., carbonate content) represent the bulk of the database and are available for most ODP Legs. Age model and paleontological information are part of the database structure, but (as of March 1998) are given only for a handful of sites. There is one general grouping (‘Paleontology’) which is divided into four searchable tables: Age Model, Paleontological Investigation, Range Table, and Species Information. Age Model information is currently (as of March 1998) available only for one hole in Leg 105 and consists of two points, the top of the drilled section and the bottom. ODP is probably planning to progressively add more detailed age model data for all ODP sites.

I used a simple, predetermined query to search the database for paleontological information and I only obtained very preliminary information, such as the name/s of the paleontologist/s who did the shipboard study, the depth in meters below seafloor of the samples analyzed, their relative stratigraphic position (e.g., middle Eocene), and the abundance and preservation of the microfossil group. As this information was available only from Leg 171 onward, legs for which no reports are published as yet, I do not know if it is planned to add more detailed paleontological information (e.g., the range charts that are available in Neptune) from the Scientific Results, once they become available.

Very basic taxonomic information is also available. For example, the search for Globorotalia truncatulinoides’ resulted in the name of the author who named the species (d’Orbigny), when the species was first described (1839), and the stratigraphic interval it is found in (Neogene).

I finally attempted to develop a customized ‘Power Query’ to search JANUS but I did not succeed. No instructions were given on how to select the various items present in the relational tables and the query routine was neither user-friendly nor intuitive.

While JANUS is undoubtedly a very valuable resource for site information and shipboard results (mainly physical properties), the preliminary search of the paleontological content of JANUS suggests that Neptune is still clearly a more valuable source of paleontological information. Although I do not see how JANUS and Neptune could be easily integrated, the two databases certainly complement each other. As shown in the studies outlined above (Lazarus et al. 1995b; Spencer-Cervato and Thierstein 1997), Neptune can be extremely helpful to biostratigraphers during ODP cruises, for example for the identification of the taxa previously recorded in a specific region during a certain time interval, thereby restricting the field of species identification to likely occurrences.

4.3. Stratigraphic research with Neptune: diachrony and hiati distribution. The field where Neptune’s potential has been already quite thoroughly exploited is stratigraphy. The chronology of Neptune’s holes has been revised several times and even if the quality of age models is quite varied (Table 2.1), it still represents the most complete and reliable data set available for stratigraphic studies. Two major groups of information have been derived from this data set, the first directly applicable to biostratigraphy, the second of a stratigraphic and paleoceanographic significance.

The goal for the first group of studies (Spencer-Cervato et al. 1993; 1994) was to determine the reliability of biostratigraphic markers, in terms of their regional versus global significance and of their synchrony or diachrony. As mentioned in Chapter 3, siliceous biostratigraphy is based on several regional calibrations of events, whilst calcareous biostratigraphy relies on a single, mainly low-latitude calibration (Berggren et al. 1995a, b). The use of the latter approach (one calibration for all holes, irrelative of their biogeographic location) implies a global synchrony of biostratigraphic events, that has actually been demonstrated only in very few cases. The first study (Spencer-Cervato et al. 1993) was aimed to calibrate several Neogene radiolarian events in the north Pacific and to study the degree of diachrony within this biogeographic region (Fig. 4.18). The projected ages of radiolarian first and last occurrences derived from the line of correlation of age/depth plots from the North Pacific have been computed from twelve North Pacific sites, and 28 radiolarian events have thereby been newly cross-calibrated to North Pacific diatom and other stratigraphy. Several of the North Pacific radiolarian events are older than in previously published equatorial Pacific calibrations (Johnson and Nigrini 1985) (Fig. 4.18), and some may be diachronous within the North Pacific. We hypothesized that these patterns may be due to complex latitudinal patterns of clinal variation in morphotypes within lineages, or to migration events from the North Pacific towards the Equator.

The second, more comprehensive study (Spencer-Cervato et al. 1994) evaluated the synchrony and diachrony of 124 commonly used Neogene biostratigraphic events in 35 globally distributed DSDP and ODP holes. Global mean age estimates based on combined biostratigraphy and magnetostratigraphy were calculated for each event. The ages’ standard deviations were used as an estimate of synchrony/diachrony. Average standard deviations for event ages by fossil group are: calcareous nannofossil first appearance datums (FADs): 0.57 m.y. (21 events), calcareous nannofossil last appearance datums (LADs): 0.60 m.y. (25 events), diatom FADs: 0.57 m.y. (7 events), diatom LADs: 0.85 m.y. (14 events), planktic foraminifera FADs: 0.88 m.y. (22 events), foraminifera LADs: 0.68 m.y. (16 events), radiolarian FADs: 0.30 m.y. (9 events), radiolarian LADs: 0.31 m.y. (10 events). 53 of the 124 events can be considered synchronous, within the resolution of the method (± two average sample spacings, i.e., 360 k.y.). The remaining diachronous events were analyzed for true patterns of diachrony and other biases. Generally, diachrony is more frequent among cosmopolitan than among endemic taxa (Fig. 4.19). Also, the precision of age calibrations decreases with increasing age. Some diachrony patterns may be due to investigator bias (see examples shown in Spencer-Cervato et al. 1994), but in general they appear to be, at least in part, real phenomena. Thus, they could provide opportunities for exploration of paleobiological processes (see for example the study on G. truncatulinoides described above, Spencer-Cervato and Thierstein 1997).

A similar study of diachrony was not attempted for Paleogene events and is not recommended either. The age control on the chronology of Paleogene sediments is poorer than what is available for Neogene sediments. Moreover, fewer sections were analyzed for magnetostratigraphy, which provides the independent control on the age models selected for the Neogene study described above. I expect that the patterns of diachrony that could be obtained for Paleogene events would be largely biased by the data set and, therefore, would not provide a scientifically sound basis for further studies.

The chronology of the 165 holes in Neptune was the subject of the third stratigraphic study originated from Neptune. It was mentioned in Chapter 3 that continuous stratigraphic sequences were very rare and that most age models were characterized by hiati. Hiati are commonly recognized in shelf sediments, but regional deep-sea hiati have also been extensively studied (e.g., Keller and Barron 1983). The reason for the interest in the timing and geographic distribution of hiati lies in the processes that cause them. A hiatus is a stratigraphic gap caused by erosion, dissolution, corrosion, nondeposition, rate of sediment supply versus dissolution (corrosion) of sediments (controlled by fluctuations in the calcite compensation depth - CCD), or shallow to deep water sediment fractionation (Berger 1970). Several studies have interpreted the occurrence of deep-water hiati in terms of changes in deep water circulation and corrosiveness (e.g., Keller and Barron 1987). Other studies have focused on the occurrence of hiati in continental shelf sediments and some authors have interpreted them within a framework of sea-level fluctuations (e.g., Vail et al. 1977; Haq et al. 1987).

For this study (Spencer-Cervato 1998) I have identified ‘hiatus events’ during the Cenozoic, based on the occurrence of individual hiati both in shelf and deep-sea sediments. The goal of the study was to test if there is a causal link between sea-level fluctuations (and climate change) and global occurrences of hiati, which are linked to oceanic circulation through a variety of complex processes. I initially attempted to reproduce the ‘global eustatic sea-level curve’ of Haq et al. (1987) with a curve of hiati distribution. This sea-level curve was constructed by the Exxon Exploration Group and is based on proprietary seismic data collected mainly on the eastern Atlantic passive continental margin. This curve has been a source of controversy since its publication, mainly because scientists had failed to reproduce it and because it was difficult to find physical mechanisms that could cause rapid sea-level fluctuations of more than 250 m, such as the ones implied in the curve. Drilling off New Jersey during ODP Leg 150X has recovered stratigraphic sequences which contain gaps that can be correlated to the ones used to construct the sea-level curve. These results (e.g., Miller et al., 1996) seem to have sedated the debates on the reliability of the sea-level curve, but the dispute on the magnitude of the fluctuations is still unresolved.

Compared to previous compilations of hiatus distribution in the DSDP stratigraphic record (e.g., Moore et al., 1978), the curve that I have obtained (Fig. 4.20c) has a better resolution (0.5 m.y.), contains more recent holes with better recovery, and is based on a more reliable and updated biochronology. Other studies (e.g., Keller and Barron 1983; Ramsay et al. 1994) instead focused on specific regions, e.g., the Indian or the Atlantic Ocean, while my study (Spencer-Cervato 1998) is of global extent.

To help in the interpretation of the record of hiati, I have estimated the paleo-water depth at which the hiati occurred and constructed three individual curves for shallow (0-2000 m), intermediate (2000-3000 m) and deep (> 3000 m) water (Fig. 4.21). The curves show that the Paleogene is characterized by few, several million-years long hiati, while the Neogene is punctuated by short, frequent hiati events (Fig. 4.20), occurring nearly synchronously in shallow and deep water sediments. The most significant Cenozoic hiatus event spans most of the Paleocene. Epoch boundaries are characterized by peaks in deepwater hiati possibly caused by an increased circulation of corrosive bottom water and sediment dissolution. The Plio-Pleistocene is characterized by a gradual decrease in the frequency of hiati. This could be caused by several factors, including the better recovery of younger sediments and therefore a lower chance of recording artificial hiati. Alternatively, this can indicate that sediment erosion and corrosion is time dependent and thus that there has been insufficient time to create hiati in the youngest sections. However, this smooth drop can also be an artifact of the time interval chosen for this analysis, which masks the high-frequency cycles of Quaternary glacio-eustatic sea level change possibly characterized by short (<0.5 m.y.) hiati, not recorded in this study.

Although some speculations were advanced on the causes of these hiatus events, their regional significance and possible causes will be the topic of future, more detailed studies. Among these, of particular interest would be the geographic distribution of hiati within ocean basins (e.g., latitudinal distribution of hiati versus latitudinal distribution of DSDP and ODP holes in the database and western versus eastern margins to identify the temporal evolution of oceanic gyre circulation) and their comparison to detailed isotopic records of deepwater circulation (e.g., Wright and Miller 1993). The depth distribution of hiati in mid-ocean and aseismic ridge sites versus continental shelf and slope sites must be also analyzed separately. These areas should be affected differently by sea level changes.

In summary, Neptune’s data have been used for five published studies of plankton evolution and stratigraphy. While the stratigraphic studies provide a quite complete overview of the potential of Neptune, the study of plankton evolution has so far been limited to biostratigraphic applications. The analysis of plankton longevity and diversity has been shown here as raw data. This is the field where Neptune’s data still have much to offer to the paleontological community. In the following section, I will explore the possibility of expanding Neptune beyond the paleontological field and will suggest possible future avenues of research based on this database.

4.4. Potential additional data for Neptune for sedimentological and paleoceanographic research

In my opinion, the potential of Neptune for future research extends considerably beyond analyses of micropaleontological data. Neptune’s chronology and relatively large number of holes are its greatest assets and they should be properly exploited. A significant step forward would be represented by the addition of sedimentological data, which would open up a whole new range of research possibilities. The expansion of Neptune would benefit the research community by providing interdisciplinary links and correlations that are at present rarely possible to scientists working on ODP material. Time pressure and poor funding force ODP-participating scientists to limit their post-cruise research to very limited, mainly isolated goals (James D. Wright, personal commun., 1998). Once their duty as sedimentologists or micropaleontologists is fulfilled and their report is submitted, scientists move quickly to the next ‘hot’ research topic, and the potential for correlations between data sets and large-scale research studies is left largely untouched. In this scenario, the opportunities provided by Neptune’s chronology and paleontological data would be greatly enhanced by other data that would allow to make large-scale, interdisciplinary (e.g., modeling) studies, or at least would provide an easily accessible source of a large amount of quality data from which to start such studies. Only very few of these studies based on deep-sea sediments are available at present (e.g., Delaney and Boyle 1988).

Among the data that should be included in Neptune, and that are consistently available at least for the more recent ODP holes, are: lithology (percentage carbonate, percentage silica); organic carbon content; physical properties (e.g., bulk density, grain density, porosity); and grain-size distribution. The field of paleoceanography would be the primary beneficiary of the combination of the existing paleontological and chronological data in Neptune with sedimentological information and physical properties data. I will mention here only a couple of the several, current research questions that are debated in the paleoceanographic community and that could be addressed with these additional data.

During the middle Miocene, important changes occurred in the climate of the Earth, an important step toward the establishment of cold polar climates and the modern climate mode characterized by glacial and interglacial cycles. These changes are documented in the oxygen isotope records (e.g., Miller et al. 1987; Fig. 4.20e), and indicate the onset of a progressive global cooling. It is not yet known what causes the abrupt shifts in climate mode that the Earth has experienced in the last million years, even though some recent evidence (Zachos et al. 1997) suggest that these shifts might have characterized the earth’s climate already since the Oligocene. It is, therefore, obvious why so many studies have focused on middle Miocene sediments and have led to the formulation of various hypotheses. Several hypotheses have linked climate changes to large-scale deepwater circulation (e.g., Shackleton et al. 1983), but the causal relationship between middle Miocene changes in deepwater circulation and the establishment of a permanent ice sheet in eastern Antarctica, is still uncertain (e.g., Kennett and Barker 1990).

Keller and Barron (1983) proposed that a "silica switch" occurred between the Atlantic and Pacific Ocean in the middle Miocene, around 15 Ma and contemporary to the 18O increase. Based on the relative abundance of siliceous sediments in nine tropical Atlantic sites, they suggested that prior to 15 Ma, Atlantic sediments were silica-rich, but that after that time, silica sedimentation switched to the Pacific Ocean. Predominantly carbonatic sediments have apparently been accumulating in the Atlantic since then. This switch would have been caused by the initiation of the Northern Component Water (NCW) circulation in the north Atlantic. However, Wright et al. (1992) have raised some concerns on the selection of data on which Keller and Barron (1983) based their hypothesis, and argue that NCW’s production began earlier (around 19 Ma) and had actually shut down during the 15 Ma 18O event. Wright et al. (1992) propose that the middle Miocene 18O increase does not correlate with deepwater circulation changes and does not represent the transition from an ice-free to an ice-house world, but is part of two or three glacial/interglacial cycles.

How could Neptune help solve this controversy? The cause of disagreement in the interpretations is the data on which Keller and Barron (1983) based their hypothesis. The possibility of modeling the results of a larger number of chronologically well constrained holes in the Atlantic and Pacific Oceans would give the ‘middle Miocene controversy’ a strong, potentially unbiased basis of data. The data that would be needed are the concentrations of carbonate and opal. These data have not been incorporated in Neptune yet, because data from different holes are potentially incompatible, due to the different methods used in their collection.

Since the beginning of the DSDP project, carbonate concentration has been a routine analysis performed on the sediments. Biogenic opal data are also available for DSDP holes, but only in the more recent ODP holes has it been estimated analytically. It would be conceptually simple to add a new table to Neptune, which would include the sample number, percentage carbonate and percentage opal. This table would be linked to the Sample Data table, which would provide an age estimate for each sample. In reality, this task is far from being trivial. The main reason for this is due to the analytical methodology used to determine the concentration of carbonate and opal. During DSDP, several different methods were used, giving results that are not comparable to each other. Routinely, percentage carbonate and opal have been estimated from smear slides, a method that has a maximum accuracy of ±10% (Hsü, Montadert et al. 1978). These data are highly subjective and not useful for rigorous quantitative studies. Keller and Barron (1983) used these counts for their "silica switch" hypothesis, on sites where the biosiliceous component was actually minimal and easily affected by dissolution of diatoms, the primary components of siliceous productivity (Wright et al. 1992). Quantitative analytical measurements of carbonate and opal would be required to map the distribution of these carbonatic and siliceous sediments in the entire ocean basins and to test the "silica switch" hypothesis.

The shipboard-based ‘carbonate bomb’ method (Müller and Gastner 1971) has been used relatively early in the DSDP on a few selected samples to provide a control on the smear slide estimates. The accuracy of this method, the most common analytical method used during DSDP cruises, is between 1 and 5%, lower for carbonate-rich sediments. In some instances, other shorebase methods (e.g., the LECO method; Hsü, Montadert et al. 1978) were compared to the results of the ‘carbonate bomb’ method and systematic differences observed. In more recent ODP Legs, a shipboard Coulometer is used to determine percent carbonate (and percent organic carbon)(e.g., Leg 121), with a precision of approx. 1%. I am not aware of a study that compares this method to the previous methods. The silica content has been quantitatively analyzed only for some ODP holes, using X-Rayx-ray fluorescence and the normative equation of Leinen (1977) (e.g., Littke et al. 1991) or, more recently, a single-step alkaline extraction method (Mortlock and Froelich 1989). The precision of the latter method is ±4%. The carbonate bomb, LECO and Coulometer estimates for carbonate and the X-ray fluorescence and alkaline extraction estimates of opal content from holes where sampling frequency is sufficient (e.g., one sample every 0.5 m.y.), could be selectively incorporated into Neptune. Some of these data are available for ODP holes from the JANUS database, but they would probably need to be added manually for the DSDP holes.

These same lithological data, combined with the paleo-water depth estimates of Spencer-Cervato (1998) (not yet included in Neptune but available at the NOAA-WDCA for Paleoclimatology Data Contr. Series #97-030: ftp://ftp.ngdc.noaa.gov/paleo/paleocean/by_contributor/spencer-cervato1998/) could be used to reconstruct the depth fluctuations of the Calcite Compensation Depth (CCD) during the Cenozoic, for which the curve by van Andel (1975) is still being used. The very smooth fluctuations of the CCD curve of van Andel (1975) do not agree with the abrupt changes that have been recently shown to characterize the earth’s climate and ocean systems in the Cenozoic. This is probably caused by the low resolution of stratigraphic studies in the 1970s. The better resolution (around 0.5 m.y. or better) of the Neptune data would allow us to refine the curve and to make it more compatible with, for example, the isotopic data currently being produced.

I briefly mentioned earlier the importance of deepwater circulation reconstruction for the understanding of the climate/ocean systems. Sediment accumulation rates and dissolution profiles can be reconstructed from physical property data and grain-size distributions, all data that are routinely produced onboard ship and that could be quite easily added to Neptune. Recent studies (Zachos et al. 1997) use a record of percent coarse fraction to demonstrate that glacial/interglacial cycles existed as far back as the late Oligocene. This parameter, combined with a high resolution 18O record from ODP Leg 154 (Ceara Rise - south Atlantic), shows a 40-k.y. periodicity, indicating a high-latitude orbital control on ice volume and temperature. This isotopic record suggests that there is an orbital control on deepwater circulation, which had not yet been shown so early probably because of the low resolution of previous studies and the paucity of deep-sea sections with high sedimentation rates and a long stratigraphic record, like the ones recovered by Leg 154 on the Ceara Rise.

The trend for future paleoceanographic studies is toward high (tens of thousands of years) resolution studies. Is Neptune ready for these studies? The answer is: not yet. The chronology of Neptune, its biggest strength and the most updated record available, is based on biostratigraphy and magnetostratigraphy, which provide an accuracy on the order of hundreds of thousands of years, at best. Oxygen isotope stratigraphy is currently the only means to obtain a better age resolution than this for Neogene sediments. Results from ODP Leg 154 provide the longest and most complete isotopic record for the whole Neogene which extends into the Oligocene (Weedon et al. 1997). A calibration of the sporadic isotopic records available for DSDP holes and the more common records from ODP holes to this recent isotopic calibration would allow us to refine the (mainly) late Neogene chronological resolution for some of the sites in Neptune. My biggest concern about this calibration is that correlations with standard isotopic records are still done by ‘wiggle matching’. Because the absolute isotopic values vary depending on the foraminifer species or sediment fraction used to obtain the record, the shape of the curve (which remains substantially the same) is used for the correlation. I am not aware of any comprehensive study that has carefully pinpointed and tabulated some of the ages of the 200+ (the total number is actually unknown: some 140 are recognized only in the Plio-Pleistocene) isotopic stages to magnetostratigraphy beyond the late Miocene (Hodell et al. 1994). This would provide fixed reference points for stratigraphic interpretations. ‘Eye-balled’ graphic correlation is, in my opinion, too inaccurate for the type of studies that it is used for (unless one can actually count back all stages starting from the Holocene) and greatly reduces the potential resolution of isotope stratigraphy.

Go to Next Section