Potential error (represented by bars in Figure 4,Figure 5,Figure 6,Figure 9,Figure 10,Figure 11) in determining extinction and origination intensities is higher for small sample sizes than for larger samples. Standing diversity of cyclostomes was relatively low in the Albian and in all post-Danian Cenozoic stages, and standing diversity of cheilostomes was very low in the Albian, Cenomanian and Turonian (Figure 2). Consequently, error bars for these are longer than for any other stages. For the times of lowest diversity enumerated above, any source of error – such as how completely the fossil record is known and random events in ‘background’ extinctions and originations – has a proportionally greater influence on calculated extinction and origination rates than during times of higher diversity. Consequently, in the discussion and interpretations that follow, we de-emphasise the earliest (Albian-Turonian) record of cheilostomes.By virtually all measures, generic extinction in the Maastrichtian was greater for both cyclostomes and cheilostomes than in any other stage during the past 100 million years. However, it is almost equalled or even exceeded (Figure 4E, F;Figure 5E,Figure 6E, F) by bryozoan generic extinction in the Danian. Error bars for all Maastrichtian and Danian extinction measures overlap, apart from measures of total bryozoan and cyclostome generic extinctions per standing diversity when stage-only genera are included (Figure 4D,Figure 5D). The proportion of stage-only genera for both cyclostomes (11 percent) and cheilostomes (14 percent) was much higher for the Maastrichtian than for any other stage; in the Danian, the 3 percent cyclostome stage-only genera is at ‘background’ level, but the 9 percent cheilostome stage-only genera is much higher than the norm of 2 to 3 percent. The greater difference in stage-only genera between Maastrichtian and Danian for cyclostomes apparently resulted in the higher Maastrichtian than Danian extinction per standing diversity per million years when stage-only genera were included. It also contributed to lack of overlap of error bars for this measure for Maastrichtian and Danian cyclostomes (Figure 5D), as well as for bryozoans as a whole (Figure 4D). Bryozoans experienced essentially identical generic extinction intensities in the Maastrichtian and Danian, except that the Maastrichtian extinction appears to have been somewhat more intense than the Danian extinction for cyclostomes. Including stage-only genera, extinctions per standing diversity were slightly higher for cyclostomes than for cheilostomes in the Maastrichtian,resulting in a smallerabsolute diversity of cyclostomes relative to cheilostomes during the Danian. There was only a slight decline in within-fauna cyclostome species richness from the Maastrichtian into the Danian, part of a longer-term trend (Lidgard et al. 1993). Conversely, there was an abrupt transition across the Maastrichtian-Danian boundary from cheilostome to cyclostome dominance of assemblage biomass as measured by relative skeletal mass (Håkansson and Thomsen 1979, 1999; McKinney et al. 1998). Therefore, the generic extinction pattern stands in sharp contrast with the abundance patterns. High generic extinction rates of bryozoans at the end of the Cretaceous have been noted previously by Viskova (1980, 1997), McKinney et al. (1998), and Sepkoski et al. (2000). This concentration of bryozoan extinctions is consistent with a K-T mass extinction, now generally attributed to impact of an extraterrestrial bolide (Alvarez et al. 1980, and many papers since). However,our data lack the precision necessary to determine whether bryozoan extinctions occurred at the end of the Maastrichtian rather than being distributed more widely through that stage. The equally high bryozoan extinction rates for the Danian must be explained by other cause(s). They apparently correlate with the loss of carbonate shelf environments across northern Europe, which hosted a high proportion of bryozoan taxa known in the Danian fossil record (Håkansson and Thomsen 1999). The widespread ‘chalk’-depositing environments were established during the Cenomanian Stage of the mid-Cretaceous (e.g., Rawson 1992; Gale 2000; Gale et al. 2000) and lasted through to the end of the Danian (e.g., Smith and Jeffery 2000). Although local bryozoan diversity and abundance were reduced essentially to zero immediately above the K-T boundary in the few complete sections studied in detail, both abundance and diversity re-built within a few decimetres of the base of the Danian section in northern Europe (Håkansson and Thomsen 1979, 1999). The subsequent disappearance of the chalk environments by the end of the Danian and correlated bryozoan extinction may possibly be explained by (1) long-term fall in sea-level (Haq et al. 1987) to a level that shut off oceanic circulation onto the shelf (e.g., Gale et al. 2000); and/or (2) development or increase of cyclonic circulation across the opening Atlantic (Parrish and Curtis 1982).Background intensities of cheilostome extinctions, calculated as extinctions per million years (Figure 6A, B), are similar for the Upper Cretaceous and Cenozoic, whereas Maastrichtian and Danian extinction rates are substantially greater (although their error bars overlap appreciably with those of several other stages). However, Maastrichtian and notably Danian extinction rates group with the highly variable earlier Cretaceous extinctions in plots of extinctions per standing diversity (Figure 6C, D) and also with extinctions per standing diversity per million years (Figure 6E, F). In contrast, for cyclostomes the measures of extinction rate are much lower for virtually all other stages than for the Maastrichtian and Danian, especially extinctions per standing diversity (Figure 5C, D). The low pre-Maastrichtian and post-Danian extinction rates likely are a good reflection of background extinction, which are more constant over time than are origination rates (e.g. Van Valen 1985;Gilinsky and Bambach 1987). Extinction rates of cheilostomes and cyclostomes during the past 100 myr have been remarkably similar, with a median (‘background’) E/D/myr of 0.009 for cyclostomes and 0.013 for cheilostomes. An arithmetic plot of cyclostome versus cheilostome extinction rates (Figure 13A) shows low correlation (r = 0.487) if all Upper Cretaceous and Cenozoic stages are considered together. However, if stages are separated into the two major time intervals, low correlation (r = 0.429) is seen for paired Upper Cretaceous extinction rates, but Cenozoic extinction rates correlate well with one another (r = 0.742). The poor correlation in the Upper Cretaceous is due largely to high variation in extinction rate of low-diversity, pre-Campanian cheilostomes, but not of cyclostomes. Sepkoski et al. (2000) determined background extinction rates for cyclostomes and cheilostomes by counting frequency of longevity of genera, grouping them into ‘bins’ of 5-million-year multiples, and taking the slope of log-linear regression for bins other than the shortest (0 to 5 million years). Their estimates were 0.31 genus/genus-million years for cyclostomes and 0.48 genus/genus-million years for cheilostomes. Background extinction rate of cyclostomes by their calculations is, therefore, 65 percent that of cheilostomes. Our results (previous paragraph) for background extinction rate calculated in a very different way—extinctions per standing diversity per million years—is remarkably similar: cyclostome extinction rate is 69 percent that of cheilostomes.Among post-Danian cheilostome and cyclostome extinction rates, the Priabonian (latest Eocene) stands anomalously high in all measures except for cheilostome extinctions per standing diversity inclusive of stage-only genera. This is probably part of the widely recognised Eocene/Oligocene extinction associated with the world change from greenhouse to icehouse conditions (Prothero 1994). We cannot resolve extinction rates in bryozoans to zone as has been done for some other taxa affected during the mid- to late Eocene extinction. However, the extinction of bryozoans appears different from the detailed extinction patterns documented for coccolithophores (Aubrey 1992) and foraminiferans (Boersma et al. 1987). While extinction patterns for these groups are collectively complex, their respective times and places of most profound extinction and displacement occurred either early or late in the middle Eocene and can be related to successive stages in the long-term cooling. Eocene extinction of echinoids also occurred in multiple phases, with the maximum diversity in the Lutetian, followed by moderate extinction at the Lutetian/Bartonian boundary and maximum extinction during the Priabonian (M. L. McKinney et al. 1992). However, extinction rates of bryozoans in the early to middle Eocene (Ypresian – Bartonian) were not above background levels but instead were intense during the Priabonian at a time when extinction rates of other taxa were declining. A possible reason for this delayed extinction of bryozoans during an extended period of global cooling is that their peak abundance occurs in shelf-depth temperate rather than tropical waters (Taylor and Allison 1998). The organisms that had been affected in the earlier phase of the extended Eocene extinction were predominantly tropical and deep-water.Cyclostome extinction rates seem to increase through the Miocene and Pliocene, culminating in a late Pliocene peak. This apparent increase in cyclostome extinctions through the Neogene, if real, finds a parallel in coral and mollusc extinctions documented in the western Atlantic and Caribbean (Stanley and Campbell 1981;Petuch 1995;Allmon et al. 1996;Jackson et al. 1993;Budd et al. 1996;Jackson and Johnson 2000). These extinctions have been attributed to a variety of environmental causes related to the closure of the Isthmus of Panama and long-distance effects of intensified glaciation.A broad range of colony growth habits has developed in both cyclostomes and cheilostomes (Lagaaij and Gautier 1965;McKinney and Jackson 1989;Hageman et al. 1998), most of which can be categorised as either encrusting or erect (a minority are free-living or have morphologies that are difficult to place, and these are not considered here). As a first step towards analysis of the ecological history of bryozoan extinctions, we examined the extinction rate of encrusting and erect genera over the past 100 million years. Some genera include both encrusting and erect colonies, and we assigned a value of 0.5 for each such genus to the tally for encrusters and also for erect forms. The most notable patterns for cyclostomes (Figure 14A) are that (1) erect forms went extinct at twice the rate of encrusters during both the Maastrichtian and Danian; and (2) the Priabonian extinction eliminated only erect genera. Indeed the Priabonian cyclostome extinction can be viewed as an intensification of the pattern seen for the Thanetian through Chattian (i.e., almost the entire Palaeogene) during which no encrusting cyclostomes are known to have gone extinct. The absence of encrusting cyclostome extinctions during the post-Danian Palaeogene and its concentration in the Neogene may, however, be a taxonomic artefact given that encrusting cyclostome genera are poorly defined. Extinction rates of encrusting and erect cheilostomes were similar during the Maastrichtian and Danian (Figure 14B), followed by disproportionally higher periods of extinctions of erect cheilostomes during the Thanetian and the Priabonian; at other times, encrusting and erect cheilostomes exhibited similar patterns of extinction rate. The substantial Palaeogene to Neogene decline in erect species from approximately 50 percent to approximately 25 percent of the bryozoan fauna (McKinney and Jackson 1989) may be due, at least in part, to the Priabonian extinction. Our data suggest contrasting extinction patterns of erect and encrusting cheilostome genera in the Pliocene and Pleistocene (Figure 14B), with high extinction rates of erect cheilostomes in the Pliocene followed by low extinction rate in the Pleistocene, and the opposite pattern in encrusting genera. These patterns for generaparallel the strong Neogene decline of erect cheilostome species in the Caribbean described by Cheetham and Jackson (1996). They noted that the decline in proportion of erect cheilostomes is in part due to more vigorous speciation of encrusting species but is largely due to preferential extinction of species in erect genera.Origination rates for bryozoan genera (Figure 9) do not show the Maastrichtian and Danian anomalies that characterise extinction rates. In general, origination rates were higher and more variable during the Late Cretaceous than during the Cenozoicand were the cause of the steep rise in overall bryozoan diversity through the Late Cretaceous (Figure 2).Cyclostome originations were high throughout the Late Cretaceous (Figure 8,Figure 10). Maastrichtian originations are the highest among these only in three of the six measures (Figure 10B-D), suggesting that Maastrichtian originations are part of the continuum of high originations during the most active period of cyclostome diversification. Following the K-T extinction, however, cyclostome originations remained very low throughout the Cenozoic by most measures, although there may have been a slight increase in originations during the Neogene (Figure 10E, F). This possible Neogene increase didnot result in an increase in cyclostome diversity (Figure 2) because it wasmatched by a slight increase in extinctions (Figure 5E, F).Cheilostome generic originations show an overall pattern of decrease through the Late Cretaceous and Cenozoic (Figure 8,Figure 11). The decline in cheilostome generic origination rate may reflect the general temporal decline in origination rates of higher taxa as a progressively greater proportion of new species are established within previously established higher level clades (cf. Flessa and Jablonski 1985), or it may mark near-saturation of the ecosystem as suggested by Sepkoski et al. (2000). Cheilostome originations during the Maastrichtian and Danian are remarkable only for the fidelity with which they fit within the long-term trend. There are three stages for which originations fall substantially below the long-term trend-line (e.g. Figure 11E, F): Upper Palaeocene (Thanetian) and both stages of the Oligocene (Rupelian, Chattian). Generic origination rates do not support Voigt's (1981) notion of the Danian as a stage of low 'creativity'. For the Cenozoic, the Danian was characterised by high origination rates, even though many of the bryozoan genera thought of as typical of the Cenozoic (e.g., Sertella, Schizoporella, Microporella) did not appear until later in the Palaeogene. The general taxonomic composition of Danian cheilostome faunas has greater similarity to Late Cretaceous faunas than to later Cenozoic faunas. This is because numerous genera that originated in the Cretaceous survived into the Danian before becoming extinct, not because of anomalously few generic originations.In contrast with the similarity in median extinction rates of cyclostomes and cheilostomes, median origination rates (O/D/myr) for cyclostomes are 0.013 and for cheilostomes are 0.028. Overall Late Cretaceous through Cenozoic origination rates of the two clades (Figure 13B) correlate better (r = 0.786) than do overall extinction rates. However, the correlation is almost entirely due to correspondence of rates during the Cretaceous (r = 0.838) when both clades were vigorously diversifying, whereas the rates are almost independent of one another (r = 0.101) during the Cenozoic. The low correlation in the Cenozoic is due to the Eocene rebound in origination rates of cheilostomes, while the cyclostomes had no corresponding rebound. Our analysis of extinction and origination of post-Palaeozoic bryozoans corroborates the inference by Sepkoski et al. (2000) that the essentially flat diversity of cyclostomes following the K-T extinction was due to sustained low origination rates rather than an increase in extinction rates. Sepkoski et al. (2000) were comparing mid-Mesozoic through Cenozoic diversities of cyclostome and cheilosotme bryozoans with coupled-logistic curves in which standing diversity of each clade suppresses origination rate (but has no effect on extinction rate) of the other. (Parameterization of variables in the calculation of the model was based on estimates of background extinction rates, equilibrium diversity, and the initial diversification rates only of the bryozoan clades.) The model, perturbed by a sudden diversity reduction simulating the K-T extinction, closely matched the actual diversity histories of cyclostomes and cheilostomes. Sepkoski et al. (2000), therefore, concluded that actual diversity history of cyclostomes and cheilostomes is consistent with competitive interference between them, which suppresses origination rates. Our analysis of origination and extinction of cyclostomes and cheilostomes broadly supports the assumptions of the Sepkoski et al. model, although, like Sepkoski et al., we note that while the history of post-Palaeozoic bryozoan diversity is consistent with a model based on competitive interference, other biological – or physical – factors may have been influential.
Cenozoic contrasts in origination rates