The Avalon Explosion: Evolution of Ediacara Morphospace

See allHide authors and affiliations

Science  04 Jan 2008:
Vol. 319, Issue 5859, pp. 81-84
DOI: 10.1126/science.1150279


Ediacara fossils [575 to 542 million years ago (Ma)] represent Earth's oldest known complex macroscopic life forms, but their morphological history is poorly understood. A comprehensive quantitative analysis of these fossils indicates that the oldest Ediacara assemblage—the Avalon assemblage (575 to 565 Ma)—already encompassed the full range of Ediacara morphospace. A comparable morphospace range was occupied by the subsequent White Sea (560 to 550 Ma) and Nama (550 to 542 Ma) assemblages, although it was populated differently. In contrast, taxonomic richness increased in the White Sea assemblage and declined in the Nama assemblage. These diversity changes, occurring while morphospace range remained relatively constant, led to inverse shifts in morphological variance. The Avalon morphospace expansion mirrors the Cambrian explosion, and both events may reflect similar underlying mechanisms.

The evolutionary history of macroscopic organisms in the late Ediacaran Period (circa 575 to 542 Ma) is regarded as a prelude to the Cambrian explosion (1). There are >270 described Ediacara species occurrences from >30 localities on several major continents (Fig. 1 and tables S1 to S3), providing an adequate data set for a quantitative analysis of Ediacara taxonomic and morphological evolution. Using cluster analysis of taxonomic composition, Waggoner recognized three Ediacara assemblages: Avalon, White Sea, and Nama (2). The Avalon assemblage (575 to 565 Ma) (3) is restricted to deep-water environments in the Avalon Province and represents an early evolutionary stage with relatively low taxonomic richness. The subsequent White Sea assemblage (560 to 550 Ma) shows a substantial diversity increase and a notable geographic expansion into Baltica (Winter Coast and adjacent area, Podolia, Finnmark, and Urals) (4, 5), Australia (6), Siberia (4), and Laurentia (7). Diversity drops again in the Nama assemblage (550 to 542 Ma) (8, 9), which includes fossils from the Kalahari Craton (10), Yangtze Platform (11, 12), Laurentia (13, 14), and Carolina Terrace (15, 16).

Fig. 1.

Ediacaran paleogeographic map and fossil localities. The Avalon assemblage is indicated by stars: 1, Charnwood Forest (England); 2, Avalon Peninsula (Newfoundland). The White Sea assemblage is indicated by squares: 3, Winter Coast and adjacent area (Russia); 4, Podolia (Ukraine); 5, Urals (Russia); 6, Finnmark (Norway); 7, Olenëk Uplift (Siberia); 8, Flinders Ranges (Australia); 9, central Australia; 10, Wernecke Mountains (Canada). The Nama assemblage is indicated by circles: 11, British Columbia (Canada); 12, Great Basin (United States); 13, southern Namibia. Baltica biota refers to Ediacara fossils of the White Sea assemblage from four localities: Winter Coast and adjacent area, Podolia, Urals, and Finnmark. Latitude lines at 30° intervals are indicated. [Modified from Waggoner (2)]

Waggoner (2) discussed three possible interpretations of the Ediacara assemblages: They may represent different evolutionary stages, biogeographic provinces, or environmental-ecological associations. These interpretations need not be mutually exclusive, because evolutionary changes could be driven by biogeographic, ecological, and environmental factors. Grazhdankin (17) argued that these assemblages represent environmental-ecological associations with little biogeographic provinciality or evolutionary change, although available geochronological data (1, 8, 9, 18, 19) indicate that the assemblages do indeed differ in age.

Previous work on the Ediacara assemblages focused on taxonomic data. We tested whether morphological patterns mirror taxonomic trends and whether there are any temporal, geographic, and environmental effects on Ediacara morphological patterns.

We recorded the presence or absence of 50 morphological characters for 272 occurrences of Ediacara species (or unnamed forms) from 60 publications and one unpublished museum sample (table S4). To maintain taphonomic uniformity, we focused on classical Ediacara assemblages and excluded possible Ediacara fossils preserved as carbonaceous compressions (20). Although the coded characters are not exhaustive and exclude some inferred anatomical structures, such as gonads and intestine of some dickinsoniid fossils (21), those characters represent the overall shape, first-order symmetry (e.g., unipolar bilateral symmetry in Charniodiscus), and central-peripheral differentiation (e.g., central stalk and primary side branches in Ediacara fronds) associated with classical Ediacara specimens. They also include such important features as stems and discoidal holdfasts. The chosen characters are easily recognizable and less likely to be altered beyond recognition by taphonomic processes, which is an important factor, considering that most Ediacara fossils are preserved as casts and molds. A coded character may not be phylogenetically homologous or functionally analogous among different taxa, but we focus on morphological evolution and make no inference on the phylogenetic homology and functional biology of the coded characters.

Our raw diversity estimates (Fig. 2A and table S3)—20, 77, and 15 genera in the Avalon, White Sea, and Nama assemblages, respectively—are broadly similar to Waggoner's estimate. To correct for uneven sampling, we used rarefaction to standardize taxonomic richness estimates. The diversity pattern observed for raw data persisted after rarefaction (Fig. 2, A and B). To further test whether taxonomic synonymy had an impact on the observed diversity pattern, we reclassified all taxa in our database on the basis of distinctive morphotypes using our character-coding system and then conducted rarefaction analysis. Again, the richness pattern remained unchanged (Fig. 2, A and B), indicating minimal impact of taxonomic synonymy. Furthermore, because the quality of fossil preservation is comparable in the three assemblages (table S3), the significant differences in taxonomic diversity are unlikely to have been an artifact of differential preservation of the assemblages.

Fig. 2.

(A) Taxonomic richness measured as the number of genera from raw data, genus-level richness standardized at 400 specimens, and the number of morphotypes standardized at 400 specimens. Error bars represent 95% confidence intervals (CIs), each estimated by 1000 independent rarefaction runs. (B) Rarefaction analysis showing relation between sampling intensity and diversity. Mean genus-level diversity and mean morphotype diversity were estimated by 1000 independent rarefaction runs. AV, Avalon; WS, White Sea; NA, Nama. (C) MDS ordination plot with convex hulls delineating the morphospace range realized by the three Ediacara assemblages. (D) Percentage of shared morphospace or genera among the three Ediacara assemblages. (E) MDS ordination plot and convex hulls of four Ediacara biotas within the White Sea assemblage: the Baltica, Flinders Ranges, Siberia, and Wernecke biotas. (F) Rarefaction analysis showing the effect of sampling intensity (number of species occurrences) on realized morphospace of the White Sea assemblage. Morphospace size (black line) and 95% CIs (red lines) estimated from 100 independent rarefaction runs. (G) MDS ordination plot and convex hulls of the Newfoundland biota of the Avalon assemblage, the Flinders Ranges biota of the White Sea assemblage, and the Namibia biota of the Nama assemblage. (H) MDS variance values (solid red line), as compared against predictions of the constant-disparity null model estimated by a Monte Carlo simulation that calculates expected variation in MDS scores if the disparity of the three assemblages was identical (the dashed red lines are the 95% CIs of the null model estimated by 1000 randomization runs). Note that two of the three assemblages are located outside the 95% CIs predicted by the null model of time-invariant disparity. MDC estimates (blue line) with 95% CIs (error bars), which were estimated separately for each of the three Ediacara assemblages by 500 balanced-bootstrap iterations.

To compare the realized morphospaces of the three Ediacara assemblages, we first used the nonparametric multidimensional scaling (MDS) method to ordinate the pooled multivariate data set into two dimensions (MDS Dim1 and Dim2), so that each species occurrence can be represented by two MDS scores rather than by 50 characters. The morphospace of each assemblage can then be visualized as a convex “hull” in a scatterplot of the two-dimensional (2D) MDS scores (22). Our results show that all three assemblages share similar morphospaces of comparable size (Fig. 2C), and this pattern persisted when the MDS ordination was fitted into three dimensions (fig. S1, A and B). Shared morphospace, calculated as the overlapping area between convex hulls, is on average 81.9% (Avalon–White Sea, 81.9%; Nama–White Sea, 84.1%; Avalon-Nama, 79.7%), as compared with 12.3% for shared genera (Avalon–White Sea, 10.3%; Nama–White Sea, 15.2%; Avalon-Nama, 11.4%) (Fig. 2D). Thus, despite substantial changes in taxonomic diversity throughout the Ediacara history, the overall size and position of the Ediacara morphospace remained markedly static.

Although the morphospace range (Fig. 2C and fig. S1, A and B) is comparable across the three assemblages, the group centroids are statistically distinct for the Avalon–White Sea and Avalon-Nama comparisons (table S5). Thus, the typical (average) morphology of the Avalon assemblage may differ from the typical morphologies of the two subsequent assemblages, perhaps reflecting the intuitive perception that the Avalon assemblage was somewhat distinct. However, the difference is minor: All pairwise distances between centroids are small (table S5), and discriminant analysis misclassifies 59.2% of species occurrences into incorrect assemblages (table S6). These results are consistent with a substantial overlap among the three morphospaces observed on ordination plots (Fig. 2C and fig. S1, A and B).

To test whether Ediacara biotas of similar age from different biogeographic provinces or paleo-latitudes have distinct morphospaces, we focused on the White Sea assemblage (i.e., the only assemblage with sufficient geographic coverage) and used MDS to ordinate the Baltica, Flinders Ranges, Siberia, and Wernecke biotas (Fig. 2E and table S2). The Baltica and Flinders Ranges biotas, likely coeval (2) but from different paleo-latitudes (23), share similar morphospaces (90.2% shared morphospace versus 41.3% shared genera) that are comparable to the morphospace of the entire White Sea assemblage [shared morphospace: Baltica–White Sea (86.9%), Flinders Ranges–White Sea (91.9%); shared genera: Baltica–White Sea (75.3%), Flinders Ranges–White Sea (44.2%)]. In contrast, the Siberia and Wernecke biotas occupy smaller morphospaces. However, these smaller morphospaces may reflect inadequate sampling, because these two biotas are represented by only 10 and 13 species occurrences, respectively. Indeed, rarefaction analysis of the White Sea assemblage suggests that at least ∼30 species occurrences are required to retain its morphospace size (Fig. 2F). In sum, the comparable morphospaces of the adequately sampled Flinders Ranges and Baltica biotas, the small distances between centroids (table S5), and poor classificatory performance of discriminant analysis (table S6) all suggest that the Ediacara morphospace may have been decoupled from paleobiogeography.

It has been argued that the distribution of Ediacara taxa was primarily controlled by paleoenvironments leoenvironments (17): Avalon-type biotas occur in deep marine habitats, Flinders Ranges–type biotas occur in shallow marine prodeltaic settings, and Nama-type biotas occur in distributary-mouth bar shoals. According to Grazhdankin (17), these three biotas represent an environmental-ecological gradient involving little evolutionary change or biogeographic provinciality. We recalculated MDS scores of the Newfoundland, Flinders Ranges, and Namibia biotas that represent these three paleoenvironments (table S2). There are substantial taxonomical differences among the three biotas [shared genera: Flinders Ranges–Newfoundland (11.8%), Newfoundland-Namibia (6.7%), Namibia–Flinders Ranges (12.8%)]. However, the percentage of shared morphospace is high [Flinders Ranges–Newfoundland (86.3%), Newfoundland-Namibia (91.7%), Namibia–Flinders Ranges (89.6%)] (Fig. 2G), and discriminant analysis suggests (tables S5 and S6) that the three groups are indistinguishable or strongly overlapping, indicating that paleoenvironments were not a major factor controlling the extent of Ediacara morphospace.

Although evolutionary change, biogeographic provinciality, and paleoenvironments might have played a role in Ediacara taxonomic evolution, they do not seem to have controlled the overall range of the realized morphospace, which appears to be invariant to notable taxonomic differences. Thus, changes in taxonomic diversity that occurred through time while morphospace range remained relatively constant should affect the internal structure of morphospace. Because of its greater taxonomic diversity, the White Sea assemblage should have a more crowded morphospace. Consequently, the morphological disparity (average morphological distances between taxa) of the White Sea assemblage should be smaller than the disparities of the Avalon and Nama assemblages.

We used three metrics—MDS variance, total character variance, and the mean dissimilarity coefficient (MDC) (2426)—to quantify morphological disparity. The MDS variance was estimated by summing the variances of MDS scores along the two MDS dimensions. As expected, the sum of the MDS variance is lower in the White Sea assemblage than the sums of the other two assemblages (Fig. 2H). Results were similar when scores based on the 3D MDS ordination were used (fig. S1C). The total character variance, calculated by summing the variances of the original 50 variables, shows a comparable outcome (fig. S1D). For binary characters, the MDC of an assemblage can be calculated as the fraction of dissimilar characters averaged across all pairwise comparisons. The MDC results are notably consistent with those based on variances, although the difference between the White Sea and Nama assemblages is statistically insignificant (Fig. 2H), perhaps as a result of low statistical power associated with the small sample size of the Nama assemblage (table S3). The inverse relation between taxonomic diversity (Fig. 2A) and morphological disparity, observed for all applied metrics (Fig. 2H and fig. S1, C and D), reflects morphospace saturation: An increase in diversity within the confines of a static morphospace inevitably reduces the average morphological distance between taxa.

Although the three Ediacara assemblages differ in taxonomic composition, their morphospaces overlap strongly and are comparable in size. Ediacara morphospace reached its maximum range already in the Avalon assemblage and was subsequently maintained in the White Sea and Nama assemblages. The morphospace was filled sparsely in the low-diversity, high-disparity Avalon and Nama assemblages but was filled densely in the high-diversity, low-disparity White Sea assemblage. Furthermore, the Ediacara morphospace ranges do not appear to have been controlled by paleobiogeography or paleoenvironments.

What might have led to the rapid morphospace expansion in the Avalon assemblage, and what might have constrained the Ediacara morphospace from further expansion or shift in the subsequent White Sea and Nama assemblages? We consider a long, undocumented period of Ediacara history before the Avalon assemblage to be unlikely. The rapid increase of morphospace at the beginning of Ediacara evolution parallels the disparity patterns of the Cambrian explosion (2729): a rapid evolution of body plans followed by taxonomic diversification within the limits of a predefined morphospace. Various environmental, ecological, and developmental factors have been proposed to explain the rapid evolution of animal body plans during the Cambrian explosion, as well as to account for post-Cambrian constraints on modifications of these basic body plans despite taxonomic diversification (30). In principle, these explanations may also be applied to the Avalon radiation. Future research should combine paleoecological, paleoenvironmental, developmental, and morphometric data to test (i) whether the Gaskiers glaciation (3), Ediacaran oxygenation (31, 32), establishment of a regulatory developmental system (33), or sophisticated ecological interactions (34) might have been the underlying drivers for the early morphological diversification of Ediacara organisms and (ii) whether the ecological saturation or developmental entrenchment might have constrained Ediacara morphospaces. Regardless of the veracity of these causative explanations, the marked parallels between the Cambrian and Avalon explosions suggest that the decoupling of taxonomic and morphological evolution is not unique to the Cambrian explosion and that the Avalon explosion represents an independent, failed experiment with an evolutionary pattern similar to that of the Cambrian explosion.

Supporting Online Material

Materials and Methods

Fig. S1

Tables S1 to S6


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article