Testing predator-driven evolution with Paleozoic crinoid arm regeneration

Regenerating arms of crinoids represent direct evidence of nonlethal attacks by predators and provide an opportunity for exploring the importance of predation through geologic time. Analysis of 11 Paleozoic crinoid Lagerstätten revealed a significant increase in arm regeneration during the Siluro-Devonian. During this interval, referred to as the Middle Paleozoic Marine Revolution, the diversity of shell-crushing predators increased, and antipredatory morphologies among invertebrate prey, such as crinoids, became more common. Crinoid armregeneration data suggest an increase in nonlethal attacks at this time and represent a causal link between those patterns, which implies an important role for predator-driven evolution.

Predation has been used to explain numerous macroevolutionary trends. For example, the hypothesis of escalation posits that prey evolve escape strategies (morphological, behavioral, or otherwise) as a consequence of interactions with their enemies. Vermeij [1][2] has argued that biological hazards, including predation, have increased through geologic time and, as a consequence, so have prey responses. This general trend has not been uniform: Several intervals of intensified escalation have been suggested, including the early Cenozoic [2], the Mesozoic [3], and the middle Paleozoic [4][5].

During the middle Paleozoic, the diversity of shell-crushing predators, especially arthropods and fishes, increased sharply and concomitantly with defensive features among mollusks, brachiopods, and crinoids, an event known as the Middle Paleozoic Marine Revolution (MPMR) [4][5]. Crinoids, which developed more spines, thicker calyx plates, and reduced viscera through this interval (Siluro-Devonian), provide an important test of escalation because the high abundance and sessile, stalked life-habit of Paleozoic crinoids made them easily accessible to predators. Escalation implies a pronounced increase in nonlethal predation during the MPMR, a hypothesis that we test here by measuring the frequency of crinoids with regenerated arms from Paleozoic Lagerstätten, or fossil deposits of exceptional preservation.

Crinoids, like all echinoderms, exhibit exceptional regenerative abilities. In addition, they possess articulations in the arms and stalk specialized for autotomy, or active shedding of body parts [6-8]. Years of observations have revealed that a variety of organisms, including starfish [9], crabs [9], sea urchins [10], and especially fishes [11-13], attack extant crinoids, which often leads to injuries. Although abiotic stimuli may also lead to trauma and subsequent regeneration, experiments, in situ observations, and functional morphology [7][8] suggest that predation is the predominant cause of autotomy and regeneration. Thus, arm-regeneration frequencies represent a measure of nonlethal predatory encounters [14-17]. We examined well-preserved crinoids, that is, those with arms largely intact. For each individual, we recorded the number of visible arms and the number of regenerating arms. Regeneratior was recognized by an abrupt change in arm diameter (Fig. 1).

Our data (Fig. 2 and table S1) show that the percentage of individuals with regenerating arms (i.e., regeneration frequency) in the early Paleozoic (Ordovician and Silurian) was <5% but increased in the middle to late Paleozoic (Devonian to Pennsylvanian) to >10%. The increase spanning the Siluro-Devonian is the only statistically significant change in the entire time series (x² P < 0.001). The increase in regeneration frequency across the Siluro-Devonian remains statistically significant (x², P < 0.001) even when (i) data are grouped by period (table S2), (ii) data are separated into two intervals by the Siluro-Devonian and (iii) individual Lagerstätten are removed one at a time ("jackknife") [18].

Paleozoic crinoids are a morphologically heterogeneous clade — size, arm number branching patterns of the arms, and armarticulation types can vary dramatically among groups. Because these factors may influence predation, regeneration, and preservation, we analyzed a morphologically more homogenous subsample, camerate crinoids [19][20]. Camerates are the most diverse and abundant crinoids found at 10 of the 11 Lagerstätten. Their absence from the Pennsylvanian sample is explained by elevated extinction in the Late Mississippian [21][22]. Camerates show the same pattern of increase in arm-regeneration frequency across the Siluro-Devonian as does the data for all crinoids combined (Fig. 2, C and D, and tables S3 and S4).

Although our data are relatively coarse and contain temporal gaps, the timing of the Siluro-Devonian increase in arm-regeneration frequency is roughly coincident with the increased diversity of shell-crushing predators as predicted by the MPMR [4][5]. Although most of our data are from eastern North America, a global extrapolation may be justified by the low variability in regeneration frequencies for a single time interval, the Silurian, when regeneration data from eastern Canada, Sweden, Great Britain, and the central United States are compared.

All crinoids used in this study inhabited the paleotropics [23] (table S1); thus, known latitudinal gradients in predation intensity [2] cannot explain the observed pattern. Although the 11 Lagerstätten do not represent a single depositional environment, the observed temporal pattern is unlikely to be an artifact of environmental differences in predator abundance and diversity, because the 4 Silurian Lagerstätten represent the full spectrum of environments yet exhibit a narrow range of regeneration frequencies (mean, 3.8%; SD, 1.1%), and the same temporal pattern emerges when comparing Lagerstätten of a single depositional type (stormdominated shelf/ramp) (table S1).

Although regeneration frequency is only a measure of nonlethal encounters, it is nonlethal encounters with predators that lead to escalation, because selection for antiprcdatory traits "can occur only if individuals in a population reproduce after they have suffered... attacks" [24-26]. Thus, the observed trend in crinoid arm-regeneration frequency implies a considerable increase in predator-driven evolution, or escalation, in the Siluro-Devonian.

• To whom correspondence should be addressed.

www.sciencemag.org/cgi/content/full/305/5689/1453/DC1

Tables S1 to S4

2 June 2004; accepted 28 July 2004

GRAPH: Fig. 2. Arm-regeneration frequencies in Paleozoic crinoids. Regeneration frequencies (%) = 100 X number of regenerating/total individuals at a locality. (A) Data for all crinoids from the 11 localities grouped into periods. (B) Data for all crinoids for each of the 11 localities. (C) Data for camerate crinoids from the 11 localities grouped by period. (D) Data for camerate crinoids for 10 of the 11 localities (the Pennsylvanian locality contained no camerate crinoids). Error bars correspond to 1 SE.

PHOTO (BLACK & WHITE): Fig. 1. Examples of regenerating arms in Paleozoic crinoids. (A) Dystactocrinus constrictus, Hall 1871, Late Ordovician, USA, USNM 93223, all arms regenerating. (B) Gennaeocrinus mourantae. Goldring 1934, Middle Devonian, Canada, USNM S4533. (C) Agaricocrinus splendens, Miller and Curley 1890, Early Mississippian, USA, USNM S8856. (D) Dimerocrinites icosidactytus, Phillips in Murchison 1839, Middle Silurian, England, USNM S8857. White arrows point to regenerating arms. Note nine regenerating arms in (D). Scale bars, 1.0 cm.