Morphological Disparity of Ammonoids and the Mark of Permian Mass Extinctions

See allHide authors and affiliations

Science  08 Oct 2004:
Vol. 306, Issue 5694, pp. 264-266
DOI: 10.1126/science.1102127


The taxonomic diversity of ammonoids, in terms of the number of taxa preserved, provides an incomplete picture of the extinction pattern during the Permian because of a strongly biased fossil record. The analysis of morphological disparity (the variety of shell shapes) is a powerful complementary tool for testing hypotheses about the selectivity of extinction and permits the recognition of three distinct patterns. First, a trend of decreasing disparity, ranging for about 30 million years, led to a minimum disparity immediately before the Permian-Triassic boundary. Second, the strongly selective Capitanian crisis fits a model of background extinction driven by standard environmental changes. Third, the end-Permian mass extinction operated as a random, nonselective sorting of morphologies, which is consistent with a catastrophic cause.

The extinctions at the close of the Paleozoic were initially understood as a relatively long process ranging for about 10 million years (My) or more, recording a progressive decline in numerous groups of marine organisms, many of which became extinct before the end-Permian (1). Most recent studies emphasize an “instantaneous”extinction event at geological scales—that is, restricted to the several thousand years bracketing the Permian-Triassic boundary (2, 3)—with a distinct mass extinction event at the end-Capitanian, 10 My earlier (2, 4).

Testing the several putative cause(s) of these events would require detailed studies of the rate, timing, and selectivity of the extinction patterns. The resulting data would permit rejection of extinction hypotheses that do not fit observed patterns. In particular, extinction selectivity can be tested using a wide array of methods (59). If we assume that the variety of forms (morphological disparity) reflects the variety of adaptive zones occupied, the analysis of disparity patterns provides a means of differentiating among various extinction models (1012).

Ammonoids are suitable subjects for analyses of extinction dynamics because they show high fluctuations in taxonomic diversity during their history and record numerous extinction events and subsequent recovery (13). We analyzed diversity dynamics of ammonoids from a distribution of 1965 species ranging from the Late Carboniferous to the Early Triassic. The data are derived from the GONIAT database, a compendium of Paleozoic species of ammonoids encompassing taxonomy, morphological data, and geographical and stratigraphical occurrences (14). The database has been extended to include Early Triassic forms and the most recent systematic developments (15).

Zhou et al. (16) interpreted variations in Late Permian ammonoid diversity (Fig. 1) as multi-episodal extinction near the stage boundary, related to sea-level fluctuations. The end-Permian is characterized by the survival of only two or three genera across the Permian-Triassic boundary (13). Pseudo-extinction of several paraphyletic taxa certainly overemphasizes the level of extinction, but the Early Triassic forms represent only three surviving superfamilies, the Medlicottiaceae and two in the Ceratitina, implying the demise of three others. However, the measured diversity pattern is strongly constrained by analytical biases and the effect of uneven sampling, preventing a confident interpretation.

Fig. 1.

Patterns of diversity for ammonoids for the interval Late Carboniferous to Early Triassic. The diversity is counted at the stage level; the stratigraphic range of taxa is taken from their first and last occurrence in the fossil record. Error bars indicate square roots of numbers of taxa. Upper panel: Number of superfamilies. Lower panel: Comparison of the number of genera (curve) and the number of regions sampled (histogram). Gray vertical bars indicate the position of mass extinctions at the end of the Capitanian and the end of the Changhsingian.

Ammonoid species and genera are taxa with a short time range, generally restricted to one stage or part of a stage. This leads to an overestimation of extinction intensity when measured at the stage level. In data for the Permian, taxonomic diversity for particular time intervals is also highly correlated with the number of geographical domains sampled (Spearman rank correlation test, P = 0.009 and P = 0.038 for genera and species richness, respectively). This indicates that measured diversity is highly sensitive to sampling effort and challenges the reliability of the temporal patterns. The crucial problem is that the critical period of the Late Permian and earliest Triassic (Griesbachian and Dienerian) suffers from a paucity of available fossiliferous sections, which can in itself explain part of the diversity patterns (17, 18).

Analysis of morphological disparity can overcome part of the sampling bias of the fossil record (19) and also complements patterns of taxonomic diversity. By focusing on shape differences, morphological disparity is relatively independent of taxonomy and allows the comparison of samples in which a variable proportion of taxa are preserved or sampled (20). Estimates of disparity consider the distribution of taxa and proportions of morphological space occupied at successive time intervals.

We constructed an empirical morphospace for Permian and Triassic ammonoids on the basis of conch coiling (whorl expansion rate, umbilical width index, whorl width index, and whorl imprint zone). Two complementary estimates of disparity were calculated: the sum of variance and the sum of range on morphospace axes. Variance measures the dispersal of forms through morphospace. It is sensitive to taxonomic choices but is statistically insensitive to the sample size, except in small samples for which uncertainty increases markedly (12). The sum of ranges estimates the amount of morphospace occupied. It is insensitive to taxonomic choices but is sensitive to sample size, the impact of which can be minimized with the use of rarefaction (20).

The temporal patterns of disparity (Fig. 2) that we found are broadly similar to the diversity curve for superfamilies (Fig. 1) (Spearman rank correlation test, P = 0.011). However, at low taxonomic levels, diversity and morphological disparity are independent (no statistical support for correlation, P = 0.718 for genera, P = 0.740 for species) and track different information. Disparity increased in the Late Carboniferous to Early Permian and remained stable during the first three Permian stages (Fig. 2). During the period from the Artinskian to the end of the Permian (∼30 My), disparity decreases, interrupted only by a brief increase in the Wordian. However, patterns of the two disparity estimates differ substantially. Fluctuations in variance are more pronounced, whereas the decreasing trend of the range exhibits more regularity. The brief disparity increase during the Wordian is related to an increase in disparity in three groups [Neoicocerataceae, Adrianitaceae, Cyclolobina (fig. S4)] that diversified during this interval (14). The end-Capitanian crisis is characterized by the loss of a large number of genera (21), and the sum of variances decreases more rapidly than does the sum of ranges. The disparity continued to decrease during the two last stages of the Permian and reached the lowest value in the Changhsingian, just before the end-Permian mass extinction. Paradoxically, the level of disparity is similar to that at the beginning of the Triassic, despite a high rate of extinction. The Triassic data show a stagnation in the variance and only a slight expansion of morphospace occupation. The post-crisis diversification of morphologies is delayed relative to the steep increase in taxonomic richness.

Fig. 2.

Patterns of morphological disparity for the ammonoid conch for the interval Late Carboniferous to Early Triassic. (A) Disparity estimated as the sum of range measures the amount of morphospace occupied. The values are rarefied to a sample size of 30 species. (B) Disparity estimated as the sum of variance approximates the mean dissimilarity between taxa. Plots are the mean value of 500 bootstrap replicates, with 90% confidence intervals as error bars.

According to the models of Foote (10), a nonselective extinction should not affect the disparity, whereas selective extinctions should modify the variance and reduce the range (or both), depending on the sensitivity to extinctions among occupants of particular adaptive zones. The progressive long-term decline during the Late Permian—demonstrated by the sum of range, and coeval to the progressive demise of superfamilies—suggests progressive erosion of the morphospace, selectively affecting the marginal morphologies (see fig. S5). This trend is inherent to the ammonoids and reflects a low rate of appearance of morphological novelties that, except for a brief interval in the Wordian, failed to compensate for ongoing background extinction during the Permian. This might be determined either by a constantly filled ecological space or by evolutionary and developmental properties of the organisms (22).

The decrease in morphological variance during the Capitanian crisis reflects a reworking of the morphospace whose margins were trimmed back, as expressed by the decrease in the sum of range, leaving surviving forms mainly clustered in its central part (21). The end-Capitanian crisis was thus selective, with most of the cases affecting the goniatitic morphologies (Adrianitaceae, Cyclolobina, Thalassocerataceae) and laterally compressed forms (Medlicottiaceae) (20). All these clades suffered during the Capitanian event or became extinct. The Wuchiapingian is distinguished from the Capitanian by the high rate of origination but absence of morphological diversification of the Ceratitina subsequent to the extinction event. Calibrated using conodont biostratigraphy, the end-Capitanian is a relatively long time interval, at least recognizable at a geological scale. Various, but not all, groups of marine organisms were affected, selectively at different times within the end-Capitanian or early Wuchiapingian (2). Although treated as occurring at the stage boundary, ammonoid extinction peaked during the earliest Wuchiapingian. The selective extinction of ammonoids and other invertebrate organisms (foraminifers, brachiopods, gastropods) can be attributed to gradual changes in environmental conditions (2, 5).

By contrast, the preservation of disparity after the end-Permian mass extinction would be expected only in a context for the nonselective extinctions of whorl morphologies. Although constructed with distinct time scale and taxonomic sampling, the index for suture line complexity (23) also provides a measure of morphological disparity of ammonoids. Its changes through time are compatible with a nonselective end-Permian extinction, showing a reduction of the range and the likely preservation of the mean value of the complexity index. Members of the Ceratitina are predominant among the survivors, but this does not reflect selectivity, as they were the taxa most likely to survive. They had the highest morphological disparity and were taxonomically the most diverse group, consisting of more than 100 species (the other clades contained fewer than 10 species each). Simulation using a rarefaction of Changhsingian diversity predicts that the survival of two groups (one being the Ceratitina) has the highest likelihood when 87% of species are randomly killed, which matches previously estimated extinction rates for the end-Permian crisis (2).

The end-Capitanian and the end-Changhsingian have been recognized as two distinct mass extinction events, the latter being marked by its intensity, but this model cannot explain all aspects of the end-Permian extinctions. Variation through time of morphological disparity suggests three independent patterns for Permian ammonoids: a long-term reduction in disparity, a high level of selective extinction at the end of the Capitanian, and a nonselective extinction at the end of the Permian. The pattern at the end of the Capitanian corresponds to a model of background extinction despite the high level of extinction. Only the end-Permian event matches the model of a “mass extinction regime”(5), arguing for a catastrophic cause consisting of a brief but major event, independent of earlier variations in diversity, with a worldwide effect and, for the most part, the nonselective demise of taxa.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article