To test competing causal hypotheses or to evaluate the effect of global change on marine microplankton, accurate estimates of the stratigraphic and temporal completeness of individual K/T boundary sections are needed. In current stratigraphic analysis the blueprint for judging complete and continuously deposited K/T boundary sequences seems to be the presence and relative thickness of a boundary clay, the presence of nannofossil zones Micula prinsii and Neobiscutum romeinii and planktic foraminiferal Zone Guembelitria cretacea (P0) followed by the Parvularugoglobigerina eugubina Zone (Pla) (Perch-Nielsen et al., 1982; Smit, 1982; Smit and Romein, 1985). By this definition, many K/T boundary sections have been judged as "relatively complete" even though the earliest Tertiary Zone P0 or the N. romeinii Zone are missing. Moreover, this method of judging chronostratigraphic completeness does not allow the recognition of the existence of intrazonal hiatuses, and may thus lead to erroneous interpretations of the nature and rate of environmental change.
Foraminiferal workers have generally examined deep-sea sections, which were believed to be more complete than shallower continental shelf sections, and found that virtually all planktic foraminiferal species disappeared simultaneously at the K/T boundary. This pattern of species extinctions is illustrated in Figure 4.2 for DSDP Site 528, which contains a relatively continuous sedimentary record with a thin laminated boundary interval (in the core catcher) (D'Hondt and Keller, 1991). Yet, as in nearly all deep-sea sections, the basal Tertiary Zone P0 and probably part of Zone Pla are missing. By judging DSDP Site 528, as well as all other low- to mid-latitude deep-sea sites that show the same sudden mass extinction and absence of the basal Tertiary Zone P0 as representing a temporally complete record, the obvious interpretation was a geologically instantaneous catastrophic event such as a bolide impact. This interpretation has even been applied to the stratigraphically more complete K/T boundary sequences at El Kef and Caravaca by Smit (1982, 1990), based on the assumption that all Cretaceous species present above the K/T boundary (except G. cretacea) are reworked (see Canudo et al., 1991, for a discussion of this problem).
Among nannofossil workers, Worsley (1974) speculated that major hiatuses characterized all deep-sea sections and shorter hiatuses were present in marine-shelf sections. This observation tended to be supported by the later discovery of the new uppermost Maastrichtian nannofossil Zone Micula prinsii and an unnamed basal Tertiary interval below the Neobiscutum romeinii subzone at El Kef by Perch-Nielsen (1979a). Despite this early recognition of an incomplete K/T boundary record in the deep-sea by both foraminiferal and nannofossil workers, few attempts at systematic chronostratigraphic analysis have been made and only for a single or a few sections (Thierstein, 1982; Herbert and D'Hondt, 1990), largely because high-resolution stratigraphic records have not been available.
Thus, what has been missing from the K/T controversy is a comprehensive chronostratigraphic and biostratigraphic data synthesis for all K/T transitions in marine depositional settings that would allow determination of the temporal completeness of each section and thereby clarify the nature of the extinction record. Such an analysis has recently been completed by MacLeod and Keller (1991a,b), who used the graphic correlation method of Shaw (1964) to compare the distribution of biostratigraphic and lithostratigraphic datums in 15 relatively complete boundary sections. These comparisons resulted in a composite estimate of 76 latest Maastrichtian to Early Paleocene (Zones P0-P1c) biostratigraphic datums for planktic foraminiferal and calcareous nannofossil species that corrects for intersequence diachroneity and allows correlations for individual sections to be made within a common chronostratigraphic model. Based on this method, analysis of 29 K/T boundary sections suggests that short, global, intrazonal hiatuses of varying duration are present in virtually