Predator-induced macroevolutionary trends in Mesozoic crinoids

Abstract

Sea urchins are a major component of recent marine communities where they exert a key role as grazers and benthic predators. However, their impact on past marine organisms, such as crinoids, is hard to infer in the fossil record. Analysis of bite mark frequencies on crinoid columnals and comprehensive genus-level diversity data provide unique insights into the importance of sea urchin predation through geologic time. These data show that over the Mesozoic, predation intensity on crinoids, as measured by bite mark frequencies on columnals, changed in step with diversity of sea urchins. Moreover, Mesozoic diversity changes in the predatory sea urchins show a positive correlation with diversity of motile crinoids and a negative correlation with diversity of sessile crinoids, consistent with a crinoid motility representing an effective escape strategy. We contend that the Mesozoic diversity history of crinoids likely represents a macroevolutionary response to changes in sea urchin predation pressure and that it may have set the stage for the recent pattern of crinoid diversity in which motile forms greatly predominate and sessile forms are restricted to deep-water refugia.

It has long been hypothesized that predator–prey interactions represent a significant driving force of evolutionary change in the history of life (1–4). However, not only is predation itself hard to detect in the fossil record, which makes it difficult to ascertain its intensity over geologic time, but macroevolutionary predictions of the hypothesis are far from simple (5–13). Recent sea urchins (Echinoidea), are known to play a key role in shallow sea ecosystems as grazers and benthic predators that can modify the distribution, abundance, and species composition of coral and algal reef communities (14–16); however, only few data have hinted at the importance of sea urchins to crinoids (17–19).

Crinoids (Crinoidea), commonly known as sea lilies or feather stars, were one of the dominant components of many shallow-sea environments through much of geologic history and a key contributor to the sedimentary record (20). Although predation by fish on crinoids and its evolutionary consequences have received the most attention (21–27), sparse data indicated that crinoids may be the prey of benthic invertebrates (28), most notably sea urchins (17–19, 29, 30). Recently it has been shown that during the Triassic, the radiation of cidaroid sea urchins capable of handling the crinoid skeleton coincided with high frequency of bite marks on crinoids likely produced by the jaw apparatus of these sea urchins (18). Because it was also during the Triassic that various modes of active and passive motility appeared among crinoids, a group that throughout its rich pre-Triassic history was almost exclusively sessile, it was argued that crinoid motility, an effective escape strategy against benthic predation, was an evolutionary response to echinoid predation (18).

The hypothesized evolutionary response of crinoids to benthic predators in the Triassic (18), however, tells us little about subsequent interactions and whether it led to any subsequent macroevolutionary consequences. Because quantitative data on the geologic history of predator–prey interactions can sometimes be gathered from trace fossils left on skeletons of prey (3) and because it has been shown that the teeth of echinoids can produce such traces (17, 18), we surveyed Mesozoic skeletons of crinoids for such bite marks (Fig. 1A and Table S1). Various traces left by predators on skeletons of their prey, such as drill holes, have often been used in a similar fashion (31). However, many complexities can plague the use of trace fossils as a predation proxy (18, 32) and recognizing the maker of the traces is perhaps most challenging. The bite traces we report were culled from among other traces on the basis of their similarity to traces found on crinoid skeletal elements retrieved from the guts and feces of extant cidaroids (17, 18). Furthermore, we collected data for stalk fragments only, as stalks are most likely to be bitten by benthic organisms, such as sea urchins, rather than fish, which have been shown to focus on crinoid arms and cups (21–25). The repeated co-occurrence of sea urchins at the localities from which crinoids with bite marks were recovered is also consistent with this interpretation.

Fig. 1.

Temporal trends in bite mark frequencies on Mesozoic motile and sessile crinoids (A). Solid line represents the mean bite mark frequencies for the six time intervals; statistical significance of changes in frequencies from one time interval to the next were evaluated using a bootstrapping procedure and are shown by asterisks (*P < 0.1 NS; **P < 0.05; ***P < 0.01), for example, the difference between LK and UK is significant at the 0.05 level (**); bite mark frequencies for motile (blue dots) and sessile (green diamonds) crinoids at localities where both taxa were found—fine dotted lines connecting motile and sessile frequencies at each locality are for visual enhancement only and the numbers correspond to localities as in Table S1; note that for all localities, bite mark frequencies are lower for motile taxa. Global Mesozoic sea urchin and crinoid (motile and sessile) diversity curves (B). Cross-correlations between changes in the average bite mark frequencies on Mesozoic crinoids and number of Mesozoic genera of sea urchins (C). Cross-correlations in C after first differencing (D). Cross-correlations between changes in the proportions of Mesozoic genera of motile crinoids and sea urchins (E). Cross-correlations in E after first differencing (F). Cross-correlations between changes in the proportions of genera of Mesozoic sessile crinoids and sea urchins (G). Cross-correlations in G after first differencing (H). Dashed lines represent least-square lines of best fit. Ma, million years ago; L, Lower; M–U, Middle–Upper; U, Upper.

Results

Our data indicate that bite mark frequencies on crinoids generally increased throughout the Mesozoic, although not with a strictly monotonic trend. Moreover, in every time bin (Fig. 1A) the frequencies of bite marks on motile crinoids were lower than those on sessile crinoids, a pattern consistent with the hypothesis based on observations of modern crinoids (17) that motility constitutes an escape strategy from benthic predation.

To test whether the documented changes in bite mark frequencies on crinoids could be a consequence of changes in the diversity of their benthic predators, we compared data on bite marks to changes in the diversity of cidaroids, camarodonts, and diadematoids (Fig. 1B), groups of regular echinoids with a strong and active jaw apparatus that were observed to feed on extant crinoids (17–19, 29, 30).

The results show a statistically significant positive correlation between trends in bite mark frequencies and sea urchin diversity (P values <0.001, Pearson r = 0.982). However, it is well known that correlations in temporal trends may be spurious (“ships that pass in the night”) and that in time series, each value is partly dependent on the previous value (value in bin t is dependent on value in bin t − 1) (33). To reduce the effect of such autocorrelation in each time series, first differencing, or comparison of changes between bins, is recommended (34). After such differencing, the correlations remain significant (P values <0.001, Pearson r = 0.989) (Fig. 1 C and D). As diversity and abundance often covary (10, 35), it is plausible that a secondary correlation between bite mark frequencies and sea urchin abundance also exists, but we have no way of independently testing that claim.

Having shown that bite mark frequencies on crinoids varied through the Mesozoic and that they were correlated with the diversity of their presumed predators, it is now possible to explore whether such changes had macroevolutionary consequences for the prey. A plausible scenario is that changes in predation pressure (inferred from bite mark frequencies) would lead to corresponding changes in the incidence of effective defenses among prey. Given that crinoid motility is an effective defense against sea urchin predation and the already established correlation between sea urchins and bite mark frequencies, two macroevolutionary patterns might be expected for sea urchins and crinoids: changes in the diversity of cidaroids, camarodonts, and diadematoids, should be correlated with changes in diversities of motile taxa, crinoids that can avoid predators both actively and passively, and anticorrelated with changes in diversities of sessile forms, crinoids permanently attached to the substrate with no obvious protection from benthic predators. These predictions were tested statistically using the genus-level diversity histories of each group obtained from the Paleobiology Database (PBDB) and other literature sources (36). Our analyses at epoch and subperiod resolution suggest strong interdependence between most observed trends (Fig. 1 E–H). Diversity of motile crinoids is positively correlated with that of sea urchins (P values <0.017, Pearson r = 0.89), whereas diversity of sessile crinoids is negatively correlated with that of sea urchins (P values = ∼0.146, Pearson r = −0.747). After differencing, changes in motile crinoid and sea urchin diversities show an even stronger positive correlation (P values <0.005, Pearson r = 0.975). Although after first differencing the relationship between sessile crinoids and sea urchins (Fig. 1 G and H) lacks statistical significance (P values = ∼0.339, Pearson r = −0.661), it should be noted that r values remain strongly negative as predicted.

Discussion

The results presented here suggest that for Mesozoic crinoids changes in the diversity of motile and sessile taxa shown in Fig. 1 most probably reflect an evolutionary response of the prey to sea urchin predation. Of course, other scenarios could be offered to explain the observed patterns. For example, one might argue for the opposite directional causation, that it was the evolutionary changes in crinoids that triggered a diversification of sea urchins leading to more intense predation and a consequent increase in the frequency of bite marks. However, given that sea urchins are omnivorous, it is more likely that their own predators, rather than prey, drove their evolution. Thus, in this case, the documented patterns are best explained as reflecting an evolutionary response of prey to their predators.

It has been argued that global diversity trends need not reflect local responses of lineages to biotic interactions and that it is at the latter, smaller scales that the evolutionary impact of such interactions is most appropriately tested (7, 8). However, our data, as well as those for several other groups (13, 37, 38), indicate that such interactions can have consequences, directly or indirectly, at higher taxonomic levels. Most compelling are cases of increasing incidence of taxa with well-developed antipredatory defenses, and it has been suggested that taxa possessing such traits may be more prone to speciation (38). Although our study does not allow us to make any claims about whether differences in speciation and/or extinction rates are the cause of changes in diversities of motile and sessile crinoids, it does show that over geologic time diversities of well-defended taxa need not change in a monotonic fashion and may be correlated with diversities of predators.

On the basis of the diversity data and the analysis of bite mark frequencies on crinoid columnals, four major phases in the Mesozoic evolutionary history of sea urchins and crinoids can be identified: (i) the Early Triassic phase that followed the near extinction of both groups, with rare crinoids and echinoids; (ii) the Middle–Late Triassic phase, when both sea urchins and motile crinoids underwent significant evolutionary radiation, whereas sessile crinoids constituted a minority; (iii) the Early Jurassic phase, when the number of sea urchins dropped, leading to a release from predation pressure on sessile crinoids, which consequently diversified; and (iv) the Middle Jurassic–Late Cretaceous phase, when diversification of sea urchins increased gradually, leading to the coevolutionary increases of diversity of motile crinoids and simultaneous decreases of sessile crinoids (starting from the Early Cretaceous). The timing of these changes is roughly coincident with the Mesozoic marine revolution (MMR) (2), although it suggests two major phases of intensification in predation-driven evolution in benthic marine communities: one in the Middle–Late Triassic and the another in Late Cretaceous times. These data are intriguingly consistent with recent suggestions that the major antipredatory innovations among benthic fauna might have occurred during two periods, the Late Triassic and the Late Cretaceous (39, 40).

To date, most evolutionary trends among crinoids connected with the MMR (2) have been ascribed to predation by fish (21–24, 26, 27). Our data suggest that benthic predation by sea urchins has also been an important (if not the main) causal driver of biological change throughout the Mesozoic, and that it may have set the stage for the recent pattern in which motile crinoids greatly predominate over sessile forms that live only at great depths (20).

Materials and Methods

Three echinoid groups with a strong and active jaw apparatus are known to interact with modern crinoids, i.e., cidaroids, camarodonts, and diadematoids (17–19, 29, 30), and genus-level data for these were extracted from the PBDB. The data were extracted using the following parameters: group names “Cidaroida” (on October 25, 2011), “Camarodonta” (on November 13, 2011), and “Diadematoida” (on November 13, 2011), and time intervals = 251 to 0 Ma. The data were then recounted per six time bins (Early Triassic, Middle–Late Triassic, Early Jurassic, Middle–Late Jurassic, Early Cretaceous, and Late Cretaceous). The genus-level data for crinoids were taken from the new Treatise on Invertebrate Paleontology, Part T, Echinodermata 2 Revised, Crinoidea, Vol 3 (36), which, at least for crinoids, is a much more comprehensive source of genus-level data than the PBDB. Motile crinoids include: (i) taxa capable with benthic locomotion, including isocrinids, holocrinids, and comatulids (except bourgueticrinids) and (ii) nektonic, planktonic, or pseudoplanktonic taxa, including pentacrinids, traumatocrinids, roveacrinids, and other microcrinoids (such as Lanternocrinus, Nasutocrinus, and Leocrinus). Sessile crinoids include taxa with a cementing or root-like mode of attachment: encrinids (except traumatocrinids), bourgueticrinids, millericrinids, cyrtocrinids, and other stalked crinoids (such as Cyclocrinus, Qingyanocrinus, Tulipacrinus, Bihaticrinus, Cratecrinus, and Taurocrinus).

Diversity data for crinoids and echinoids were subjected to a runs test (“linear models”) on PAST 2.02 to find P and Pearson r values (41).

Bite marks were counted on every affected skeletal element, such that for articulated columnal ossicles (= pluricolumnals), each columnal within a pluricolumnal was counted separately.

The frequencies of bite marks are higher on sessile taxa in all five time bins in which both sessile and motile taxa coexisted (Table S1). This pattern is consistent with the hypothesis that motile taxa should be less prone to benthic predation because the null model, that the probability of either one being higher in any time bin is equal (P = q = 0.5), can be rejected at P ∼0.05.

A test of statistical significance of differences in bite mark frequencies between adjacent time bins was done using a bootstrapping procedure that involved calculating 1,000 average bite mark frequencies for each bin by sampling each bin with replacement and comparing the distribution of bootstrapped averages in adjacent bins. All differences in bite mark frequencies between adjacent bins were significant at P < 0.05, with the exception of Middle–Upper Jurassic to Lower Cretaceous, which was marginally significant (P < 0.1). The correlations for Fig. 1 C and D are statistically significant (P < 0.05) regardless of whether mean, median, or logit transform is used.