BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofitpublishers, academic institutions, research libraries, and research funders in the common goal of maximizing access tocritical research.

Temporal trends of predation resistance in Paleozoic crinoidarm branching morphologiesAuthor(s): V. J. Syverson and Tomasz K. BaumillerSource: Paleobiology, 40(3):417-427. 2014.Published By: The Paleontological SocietyDOI: http://dx.doi.org/10.1666/13063URL: http://www.bioone.org/doi/full/10.1666/13063

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in thebiological, ecological, and environmental sciences. BioOne provides a sustainable onlineplatform for over 170 journals and books published by nonprofit societies, associations,museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associated contentindicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, and non-commercialuse. Commercial inquiries or rights and permissions requests should be directed to theindividual publisher as copyright holder.

Temporal trends of predation resistance in Paleozoic crinoid armbranching morphologies

V. J. Syverson and Tomasz K. Baumiller

Abstract.— The rise of durophagous predators during the Paleozoic represents an ecological constraintimposed on sessile marine fauna. In crinoids, it has been suggested that increasing predation pressuredrove the spread of adaptations against predation. Damage to a crinoid’s arms from nonlethalpredation varies as a function of arm branching pattern. Here, using a metric for resilience to predation(‘‘expected arm loss,’’ EAL), we test the hypothesis that the increase in predation led to more predation-resistant arm branching patterns (lower EAL) among Paleozoic crinoids. EAL was computed for 230genera of Paleozoic crinoids and analyzed with respect to taxonomy and time. The results showsignificant variability among taxa. Camerates, especially monobathrids, display a pattern ofincreasingly convergent and predation-resistant arm morphologies from the Ordovician through theDevonian, with no significant change during the Mississippian. In contrast, the mean EAL amongcladids follows no overall trend through the Paleozoic. Regenerating arms are known to be significantlymore common in camerates than in other Paleozoic taxa; if regeneration is taken as a proxy fornonlethal interactions with durophagous predators, this indicates that nonlethal predation occurredmore often among camerates throughout the Early and Middle Paleozoic. In addition, frequency ofinjury among camerates is inversely correlated with EAL and positively correlated with infestation byparasitic snails. From this we conclude that decreasing EAL signals a selective pressure in favor ofresistance to grazing predation in camerates but not in other subclasses before the Mississippian, withan apparent relaxation in this constraint after the late Devonian extinctions.

V. J. Syverson and Tomasz K. Baumiller. Museum of Paleontology, University of Michigan, Ann Arbor,Michigan 48104, U.S.A. E-mail: vjsyvers@umich.edu and tomaszb@umich.edu

Accepted: 9 December 2013Published online: 7 May 2014Supplemental materials deposited at Dryad: doi:10.5061/dryad.5h45c

Introduction and Background

Stalked crinoids live an exposed and pri-marily sessile lifestyle, vulnerable to preda-tors. However, their regenerative capacitiesmean that they can recover from most damagethat is not fatal. Such nonlethal predation isthought to be frequent in modern crinoids asinferred from both truncated arms and absentor regenerating visceral masses, and predatorshave been observed carrying away arms(Mladenov 1983; Meyer et al. 1984; Meyer1985; Schneider 1988; Nichols 1994). Althoughthere are other potential sources of arm lossresulting in regrowth, including abiotic trau-ma, physiological stress, and normal ontoge-ny, most damage in modern crinoids isthought to result from biotic interactions(Mladenov 1983; Meyer 1985; Lawrence andVasquez 1996). In Paleozoic crinoids, theexistence of nonlethal predation is attestedby the presence of regenerating arms in fossilspecimens (Oji 2001; Baumiller and Gahn

2004; Gahn and Baumiller 2005), and theidentity of predators is suggested by placo-derm-like bite marks (Gorzelak et al. 2011).

A wide variety of features in crinoids havebeen described as possible adaptations topredation. These include (1) behavioral andmobility-related adaptations, such as noctur-nal activity, semicryptic habit, swimming andcrawling (Meyer and Macurda 1977; Vermeij1977), and deep habitat (Bottjer and Jablonski1988); (2) biochemical defenses, such asunpalatability (Rideout et al. 1979; McClintocket al. 1999) and aposematic coloration (Law-rence 2009); (3) physical defenses such as thickor spiny calycal plates (Signor and Brett 1984)or dense spiny pinnules proximal to the oralsurface (Meyer 1985); and (4) optimizationssuch as locating the gonads far away frompotentially fatal areas (Lane 1984), ontogeneticloss or autotomy of the stalk (Baumiller 2008;Baumiller et al. 2008; Janevski and Baumiller2010), and autotomy and autotomy-related

� 2014 The Paleontological Society. All rights reserved. 0094-8373/14/4003-00/$1.00

Paleobiology, 40(3), 2014, pp. 417–427DOI: 10.1666/13063

optimizations of the arms (Oji and Okamoto1994). The last of these is the focus of the workpresented herein.

Although most studies of functional mor-phology in crinoid arms have tended to focuson improvement of feeding ability (Cowen1981; Kammer and Ausich 1987; Baumiller1993; Brower 2006), Oji and Okamoto (1994)observed that there are arm branching pat-terns that reduce the damage sustained whenarms are lost, which may not necessarilycoincide with optimal feeding strategies. Theydescribed two optima in the space of possiblearm forms, given that the loss of even aportion of a food-gathering appendage isdetrimental to the organism even if it canregenerate, which they called the ‘‘harvesting’’and ‘‘anti-predation’’ paradigms. In the for-mer, for a certain total length of arms in asymmetrical, planar organism, food gatheringefficiency is thought to be maximized whenthe branches are spaced uniformly throughoutthe crown; in the latter, arms branch very closeto the base so as to minimize loss when thearm is autotomized as near as possible to thepoint of injury, assuming the probability ofinjury to be uniform along the arm length.Among post-Paleozoic crinoids, they foundthat anti-predatory morphologies have in-creased in frequency since the Jurassic to nearuniversality among modern crinoids; this,they suggest, represents an adaptation topost-Paleozoic predators.

Although specialized arm autotomy articu-lations (syzygies/cryptosyzygies) may nothave been present in Paleozoic crinoids (Oji2001), the same morphological optimizationsapply to arm loss via predator attack. Thefrequency of regenerating arms has beenfound to change over the Paleozoic (Baumillerand Gahn 2004), suggesting that predationpressure leading to arm loss may have variedalso. Changes in the diversity and composi-tion of predators in the Paleozoic have alsobeen recognized. For example, the Devonianhas been identified as a time of intensifieddurophagous predation (Signor and Brett1984; Bambach 1999; Dahl 2010), and morerecently the end-Devonian Hangenberg ex-tinction was recognized as a period of highturnover among predatory fishes (Sallan and

Coates 2010). If some of these changes affectedthe intensity of nonlethal predation on cri-noids, changes in the frequencies of morepredation-resistant arm morphologies wouldbe expected. We therefore chose to explorehow crinoid arm morphologies changed dur-ing the Paleozoic, specifically focusing onchanges in frequency of morphologies resis-tant to partial predation.

Materials and Methods

Expected Arm Loss.—In order to quantifymorphological resistance to nonlethal preda-tion, we used the ‘‘expected arm loss’’ metricof Oji and Okamoto (1994). As describedabove and discussed in detail in that paper,the arm branching morphologies of crinoidsare not all equivalent in terms of the propor-tion of arm loss during a nonlethal encounterwith a predator: some branching morpholo-gies result in a smaller proportion of arm lossduring an encounter than do others. Theirmodel makes the following assumptions: (1)individuals are pentaradially symmetrical; (2)all nonlethal attacks have an equal probabilityof severing the arm at any point along itslength; and (3) the arm is lost completelyabove the point of attack and unaffectedbelow it. An arm, for these purposes, isdefined as all brachials proceeding from asingle radial; the expected arm loss cantherefore be compared across taxa withoutregard for the number of free arms, because allindividuals in the sample have five radials.

We omit from our analysis all genera forwhich the first assumption does not hold. Thesecond assumption is a simplification, but onethat can be easily relaxed. An equal probabil-ity model is most neutral, as it assumes noknowledge of the predator’s preferences;however, if attacks are known to be concen-trated on any specific part of the arm, themodel can accommodate such alternativedistributions of probability. Most conditionsthat would violate this assumption wouldlead to the expectation of the arms beingsevered closer to the base, and the effect onEAL of branching closer to the base istherefore increased. For instance, predatorseating the crinoid’s arms might preferentiallybite off arms near their bases; or if predators

418 V. J. SYVERSON AND TOMASZ K. BAUMILLER

were targeting parasites, the probability ofattack would be elevated at the locationspreferred by the parasites, which we mightexpect to be located near the mouth forpurposes of stealing food or excreta. The thirdassumption is likely to be valid because mostPaleozoic crinoids had undifferentiated armarticulations. Only among a few, the advancedcladids and some camerates, were the mostproximal articulations different from all oth-ers, and even among those taxa none havebeen recognized with the specialized articula-tions for autotomy characteristic of moderncrinoids (Oji 2001), although the phenomenonhas not been fully explored. Thus, whereas inmodern crinoids failure occurs at these spe-cialized articulations, in Paleozoic crinoids weassume that it would occur directly at thedamaged articulation, because there was nopreferred place of failure.

Given the above assumptions, expected armloss (EAL) is defined as the expected value forthe proportion of a single arm lost in anysingle attack from a predator:

EAL ¼X

brachitaxes

length of segment i

total length

� �

3length above segment i

total length

� �

This gives an estimate of how susceptiblethe animal is to such damage. Lower EALindicates more predator-resistant morphology.

As an example, consider a crinoid with fivesimple bifurcating arms that divide halfwayup their length, as illustrated in Figure 1A. Wewill illustrate the computation of EAL for onearm of this crinoid step by step.

1. First, consider an injury that occurs on oneof the two free arm segments, above thenode (the branching point). This segmentmakes up 1/3 of the total arm length,which given assumption 2 (evenly-distrib-uted probability of attack) means that theprobability of injury on that segment is 1/3. Given assumptions 2 and 3, on averagesuch an injury would result in the loss of 1/2 of that segment. Thus an injury on one ofthe two free arm segments results in the

expected arm loss of [(1/3) * (1/2)]¼ 1/6 ofthe total arm length.

2. Second, consider an injury that occursbelow the node, on the lower segment.The probability that this segment will beinjured is 1/3, equal to that of the othertwo segments, because they are all of equallength. However, a strike below the nodeleads to the loss both of 1/2 of the lowersegment (1/6 of total arm length) and ofthe two segments above the node (2/3 oftotal arm length), in total 5/6 of armlength.

3. To calculate EAL for this arm branchingpattern, we add up the expected losses foran injury on each segment; i.e., for theupper left segment [(1/3) * (1/6)], for theupper right segment [(1/3) * (1/6)], and forthe lower segment [(1/3) * (5/6)]. The EALfor this branching style is therefore 7/18:2[(1/3) * (1/6)] þ 1[(1/3) * (5/6)].

For comparison, an arm that bifurcates onceat the base (Fig. 1B) has an EAL of 2[(1/2) *(1/4)]¼1/4 ; bifurcating twice at the base (Fig.1C) halves that to 4[(1/4) * (1/8)] ¼ 1/8. Thecomputation of EAL for real crinoids, such asa typical camerate (Fig. 1D) and a typicalcladid (Fig. 1E), is the same. In general, EALdecreases (indicating less vulnerability topredation) when the number of free arms isincreased or when they branch closer to thebase, as shown in Figure 2.

Data.—In order to characterize changes inthe prevalence of predation-resistant morphol-ogy, we measured the arms and calculated theEAL for crinoid genera ranging across thePaleozoic. Individual specimens were chosenfor the presence of at least one arm structurereasonably complete and consistent with thegenus description, and for presence in Web-ster’s compendium of Paleozoic crinoid gen-era (Webster 2003). The final sample includeda total of 229 genera; of these, there were 74Camerata and 139 Cladida, with the remain-ing 34 distributed among the Flexibilia andDisparida. Names and EAL for all genera inthis study, with origin and extinction datestaken from Webster (2003), can be found inSupplementary Table A.

IMPACT OF PREDATION ON CRINOID ARMS 419

FIGURE 1. Measurements and calculation of expected arm loss (EAL). A–C, Three simplified crinoid arms. If the animal in A isattacked by a predator that bites off a single arm at a random point, for each of the three segments the probability of the injuryoccurring on that segment is 1/3. If the injury occurs on one of the two free arm segments, the arm loses on average 1/6 of itslength; if it occurs on the lower segment, it loses 5/6 of its length. The EAL is the sum over all segments: 2[(1/3) * (1/6)]þ1[(1/3) * (5/6)]¼7/18¼0.39. By the same reasoning the arm in B has half its length in each segment and two segments, for an EAL of0.25, and that in C has an EAL of 0.125. D, E, Photographs and calculations for two typical specimens. D, Abatocrinus (acamerate). Brachials incorporated in cup, indicated by dashed line, are counted as zero length, so

EAL ¼ L1*L12ð Þþ L2*

L22ð Þþ L3*

L32

� �þ L4*

L42ð Þ

L1þL2þL3þL4:

E, Blothrocrinus (a cladid). Stars indicate broken free arms.

EAL ¼ L1*L12 þL2þL3þL4þL5þL6þL7ð Þð Þþ L2*

L22 þL3þL4ð Þð Þþ L3*

L32

� �þ L4*

L42ð Þþ L5*

L52 þL6þL7

� �� �þ L6*

L62

� �þ L7*

L72ð Þ

L1þL2þL3þL4þL5þL6þL7:

420 V. J. SYVERSON AND TOMASZ K. BAUMILLER

A total of 198 of the 230 genera weremeasured from plates in Volume T of theTreatise on Invertebrate Paleontology (Mooreet al. 1978). An additional 32 photographs and31 physical specimens from the private collec-tion of Joseph M. Koniecki (www.crinus.info)were measured to give estimates of within-genus variability, fill in the intervals for whichfew good specimens were available in theTreatise, and assess possible biases due to theflattening of arm structures to two-dimension-al images. Photographs and plates weremeasured using Adobe Illustrator; real speci-mens were measured using a flexible wire anda ruler. Data were recorded in a format thatpreserved the length, relationship, and state ofpreservation of all brachitaxes. Brachials in-corporated in the calyx were recorded as zerolength. The EAL for each genus was calculatedfrom the measured arm structure as describedabove. These data are included in the supple-mentary material.

We tested for biases introduced by the useof plates and photographs instead of physicalspecimens, preservation quality, specimensize, and inconsistency between differentcollections. Results of tests for bias are givenin Table 1. There was no significant differencebetween measurements of EAL obtained fromTIP plates and those from modern photo-graphs of the private collection, nor was therea significant difference between measure-ments obtained from those photographs and

the physical specimens themselves. We con-cluded that EAL is robust to differencesbetween collections and that no significantbias exists in measurements taken fromphotographs or plates relative to actualspecimens. The difference between specimenswith intact arms and those in which thelongest free arm was broken was borderline-significant, but the magnitude of the effectwas small. Within the final total of 229 genera,our sampling is reasonably reflective ofoverall Paleozoic crinoid generic diversity asdescribed by Webster (2003); for each time bin,about one-quarter of the genera in thatdatabase are present in our sample (l ¼ 0.26,r2 ¼ 0.08). For further information on sam-pling, see Supplementary Figure B.

In order to determine the robustness of theEAL measure, we calculated standard errorsfor single-species and single-genus collections.The single-species data comprise multipleexamples of all five rays from specimens ofthe camerate Amphoracrinus viminalis, whosearm branching is described as ‘‘highly vari-able,’’ from the early Tournasian MeadvilleShale of Ohio (Ausich and Roeser 2012: p.492). A bootstrap analysis of the A. viminalisdata was conducted by calculating the EALfor each figured ray, recombining them 1000times into ‘‘individuals’’ with five rays each,and taking the mean EAL for each of them.Standard error for these data was 0.005;because A. viminalis has unusually highvariability in arm branching, this is probablynear the upper limit for within-species varia-tion. Bootstrap standard error for single-genus

FIGURE 2. EAL as a function of node location for differentnumbers of free arms. The number of free arms and thelocations at which they branch govern the EAL value for agiven arm; the minimum value is therefore infinitesimaland the maximum is 0.5. Note that EAL decreases as thenumber of free arms increases and as nodes shift towardthe base.

TABLE 1. Tests for bias due to data source and preservationquality (Mann-Whitney U-test; H0 ¼ no difference/nocorrelation); significant results indicated by boldface.Measurements from photographs were tested againstmeasurements taken directly on the specimens of whichthe photographs were taken. Brokenness was based onwhether the longest such length was on a broken arm;difference between broken and unbroken specimens wasborderline significant, but the separation was not large(overlap between the populations is complete at U/Umax¼0.5).

Ratio p-value U/Umax

Treatise ~ collection photos 198:32 0.3345Photographs ~ specimens 31:31 0.9607Broken ~ unbroken 79:150 0.0566 0.4232

IMPACT OF PREDATION ON CRINOID ARMS 421

collections of Arthroacantha (ten specimens)and the cladid Cupulocrinus (28 specimens)taken from Mr. Koniecki’s private collectionwere, respectively, 0.008 and 0.005, as com-pared to mean within-time-bin standard devi-ations of 0.05 and 0.1 for camerates andcladids, respectively. We conclude that with-in-species and within-genus EAL variability islow compared to differences between genera.The standard error for all genus EAL values inthe data set was set to 0.006, the mean ofArthroacantha and Cupulocrinus. The valueswere grouped into time bins, and means of allgenera present within each bin were tested forstatistically significant correlation with time.

Results

Our results are summarized in Figure 3.There is no significant temporal trend in meanEAL for all crinoids over the Paleozoic, asshown in Figure 3A. However, when the twolargest Paleozoic crinoid clades, cameratesand cladids, were analyzed separately (Fig.3B), a strikingly different pattern emerged:mean camerate EAL exhibits a significantdownward trend over the Paleozoic, whilethe cladids show no net trend. The significantdecrease in camerate EAL is not strictlymonotonic; the steep decrease in the early tomid Paleozoic is followed by an interval oflow, but stable, EAL in later Paleozoic. Asdiscussed above, EAL is governed by twoproperties of the arm: the number of free armsand the height at which they branch (Fig. 2).Time-bin means of EAL and number of freearms in camerates show no significant corre-lation (p¼ 0.2), leading us to conclude that theaforementioned decrease in camerate EALoccurred via an increase in the number ofcamerate taxa with arms branching proximalto the calyx, rather than an increase in thenumber of free arms.

Qualitatively, these patterns are robust withregard to bin size and evenness (see Supple-mentary Fig. A). The p-values given in Table 2were calculated using the ICS epoch time bins,but they do not change significantly whendifferent bin sizes are used. All show aconsistently decreasing value of EAL amongcamerates and fluctuating values of EALamong cladids during the Paleozoic.

Discussion

We have argued that a lower EAL is moreadaptive in situations where nonlethal preda-tors represent a substantial burden on cri-noids. Our results indicate that EAL declinedsignificantly in one major crinoid clade, thecamerates, but not in the other, the cladids. Ifour adaptive hypothesis is correct, we wouldexpect nonlethal predation pressure to behigher for camerates than for non-camerates.To test this, we need an independent measureof predation pressure.

Nonlethal predation intensity on crinoidshas generally been estimated from the fre-quency of injured individuals (e.g. Meyer1985; Schneider 1988). Baumiller and Gahn(Baumiller and Gahn 2004; Gahn and Baumil-ler 2010) extended this approach to Paleozoiccrinoids by counting the proportion of injuredand regenerating crinoids in numerous Lager-statten from the Ordovician through Pennsyl-vanian; these data are given in Figure 4 andTable 3. Camerates, the dominant group intheir samples, were found to be regeneratingsignificantly more often than expected (bino-mial p , 0.01). Cladids, the second mostabundant taxon, were injured significantly lessoften than expected (binomial p , 0.01). Whencompared directly, frequency of injured cam-erates is significantly higher than that ofinjured cladids (v2 p , 0.0001). Additionally,in each period from the Ordovician throughthe Mississippian for which Baumiller andGahn (2004) were able to gather data oncamerate injuries, (1) the incidence was from2.5 to 12 times higher than among non-camerates, and (2) the temporal trend in injuryfrequencies exhibits a significant (q¼�0.8, p¼0.005) correlation with EAL, as shown inFigure 5. Thus, if frequencies of injuries areaccepted as a proxy for predation, theseresults suggest that camerates were underheavier predation pressure than expected andsignificantly greater pressure than cladids.(Injuries to disparids, the third most abundanttaxon, were also significantly lower thanexpected.)

In order to use the number of visibly injuredand regenerating individuals as a proxy forintensity of partial predation, following Gahn

422 V. J. SYVERSON AND TOMASZ K. BAUMILLER

TABLE 2. p-values of Spearman rank-order tests for correlation of crinoid EALs (logit) with age and specimen size,calculated on the basis of maximum arm length from radial to tip (H0¼ no correlation). Significant results indicated byboldface. Correlation of EAL with age was significant only for camerates; when split according to the apparent change intrend in the Late Devonian, the correlation was even stronger and more significant for camerates during the first half ofthe Paleozoic, and not significant during the second half. Correlation of EAL with size was significant only for cladids.

All E Ord – L Dev L Dev – L Perm

All crinoids ~ age 0.9160 0.8648 0.1361Camerates ~ age 4.6310�6 (q ¼ 0.90) 2.2310�16 (q ¼ 0.94) 0.7489Cladids ~ age 0.2512 0.2629 0.136All crinoids ~ size 0.3641Camerates ~ size 0.9422Cladids ~ size 4.3310�4 (q ¼ �0.29)

FIGURE 3. Mean EAL values through time; lower values indicate more predator-resistant morphologies. A, All genera insample. Numbers along bottom axis indicate sample size in time bin. B, Genera split by subclass. Numbers at top andbottom indicate respectively the number of cladids and the number of camerates in each bin. Error bars indicate 1bootstrapped standard deviation. Note that cladid and camerate values diverge by the Devonian. Neither the wholesample nor the cladid subsample displays a clear trend over time, whereas camerate EAL decreases up to the lateDevonian and stays uniformly low thereafter.

IMPACT OF PREDATION ON CRINOID ARMS 423

and Baumiller (2004, 2005, 2010), we mustmake two assumptions: (1) a consistentproportional regeneration rate across the taxabeing compared, and (2) a very low ratio offatal to nonfatal injuries. If both of theseassumptions hold, then the number of indi-viduals with visibly regenerating arms accu-rately reflects the rate of injury, and thereforethe rate of predator-prey interactions. For amore complete discussion of this problem, seeBaumiller (2013). Additionally, the number offree arms might conceivably have an influenceon either injury frequency or regenerationrate. For those genera present in both this data

set and that of Baumiller and Gahn (2004),though, we find no correlation (p . 0.4)between the number of free arms and theproportion of individuals regenerating at leastone arm.

These results on arm regeneration frequen-cies correspond to what one would expect ifnonlethal predation were the factor drivingthe evolutionary response of camerate arms.We suggest that the changes in arm branchingmorphology were indeed driven by predatorpressure that was selectively greater oncamerates, and that lower EAL is an anti-predatory adaptation in early Paleozoic cam-erates, just as in Mesozoic crinoids. Weconclude that the changes in arm branchingmorphology were indeed driven by predatorpressure that was selectively greater oncamerates, and that lower EAL is an anti-predatory adaptation in early Paleozoic cam-erates, just as in Mesozoic crinoids.

Ecological Correlates of Predation.—If, as wehave suggested here, camerate trends in armbranching morphologies leading to lowervalues of EAL were a consequence of signif-icantly higher frequencies of injuries, we areleft with the question of why this would betrue for camerates and not other crinoids. Onepossibility is that predation pressure wasconstant on all crinoids, but camerates suf-

FIGURE 4. Proportion of regenerating arms by period forall camerates and cladids. Data from Gahn and Baumiller(2004).

TABLE 3. Frequency of regenerating arms and ofinfestation in camerates and cladids. Significant resultsindicated by boldface. Regenerating arms are significantlymore common in camerates, and less common in cladids,than expectation. If regenerating arms are accepted as aproxy for injury by predators, then this indicates thatpredators preferred camerates over cladids as prey.Infesters show a significant preference for cameratesover all non-camerates and over cladids in particular.Data from Baumiller and Gahn (2004).

Camerate Cladid p (v2)

Regenerating 160 31 6.65 3 10�8

Total 1381 652Infested (individuals) 17 5 0.0165Total (individuals) 44 35Infested (genera) 17 5 6.28 3 10�4

Total (genera) 44 53

FIGURE 5. Arm regeneration and platyceratid infestationfrequencies in camerates (A) compared with camerateEAL values (B). Period time bins.

424 V. J. SYVERSON AND TOMASZ K. BAUMILLER

fered lower mortalities, i.e., were better able tosurvive predatory attacks. At present we haveno data to evaluate this hypothesis; nomorphological, physiological, or behavioralfeatures are known or suspected to makecamerates more resilient. An alternative ex-planation is that grazing pressure was higheron camerates. Grazing predation on epibiontshas been hypothesized as an explanation forwhy modern fish have been observed to biteoff arms of crinoids and spit them out: thepredators’ main targets could be the numer-ous and diverse parasites, commensals, andepibionts instead of the crinoids’ distastefularms (Meyer 1985; Brett 2003; Baumiller 2008).The extreme cryptic coloration of many ofthese epibionts, camouflaging them againstthe crinoids’ often vivid coloration, furthersuggests that they are subject to selection fromvisual predators such as fish (Hempson andGriffiths 2008). Were crinoid epibionts thetargets of Paleozoic predators and, if so, whywould camerates experience greater intensityof this type of interaction?

A possible answer is offered by the findingthat parasitic platyceratid snails prefer cam-erate hosts (Gahn and Baumiller 2006). Platy-ceratid infestation occurs overwhelmingly incamerates, and their frequency declines alongwith that of camerates during the late Paleo-zoic, although their preference is not sensitiveto time or correlated to EAL; see Figure 5 andTable 4. The presence of parasitic platycer-atids, which position themselves on the oralsurface of the calyx, might draw the attentionof predators, perhaps resulting in incidentaldamage to the arms (Brett and Walker 2002;Brett 2003; Brett et al. 2004).

In order to investigate this further, wereanalyzed data from the Paleozoic Lager-statte reported by Gahn and Baumiller (Bau-

miller and Gahn 2004; Gahn and Baumiller2005, 2006). All crinoid genera from thosestudies were categorized as platyceratid hostsor not; as injured (with regenerating arms) oruninjured; and as camerates, cladids, dispar-ids, or flexibles. A chi-squared test for thetaxonomic preference of infestation among thegenera reported by Baumiller and Gahn(2004), given in Table 3, shows that camerategenera are significantly (p , 0.001) more likelyto be infested than non-camerates and cladids(p , 0.05). Injuries are significantly morecommon in genera known to be hosts thanin genera that have not been recognized ashosts (Table 4), regardless of whether onecounts genera (p , 0.01) or individualspecimens belonging to a given genus (p ,

0.001). This is consistent with the hypothesisof platyceratid targeting (Brett 2003), though itis also possible that parasites and predatorsboth targeted the same camerate taxa foranother reason, such as food-gathering ability.

The downward trend in camerate EALreaches its minimum in the Late Devonian,after which camerates show little change inEAL. This may be due to a natural minimumvalue to the adaptation: the Devonian formswith the lowest values of EAL have many freearms that branch at the base, and furtherreduction in EAL could be achieved only byadding more arms. It is possible that there issome maximum number of arms past whichcrowding reduces filtering capacity, or thatmultiple closely packed adjacent arms can bebitten off by a predator all at once, obviatingthe advantage of having more.

Alternatively, an ecological change, such asan extinction, may have altered the selectivepressure imposed by predators. It has beenpostulated that the taxonomic turnoveramong fishes during the Hangenberg extinc-

TABLE 4. Frequency of regeneration in genera with infesting platyceratids versus those without. Significant resultsindicated by boldface. Genera known to be hosts are significantly more likely to be injured than those on which noparasites have been found. Data from Baumiller and Gahn (2004).

Infested Uninfested p (v2)

Regenerating (genera) 15 26 0.00817Total (genera) 23 76Regenerating (individuals) 109 89 3.01 3 10�8

Total (individuals) 869 1488

IMPACT OF PREDATION ON CRINOID ARMS 425

tion led to a change in the dominant mode ofdurophagous predation. The dominant Devo-nian durophagous fishes, placoderms, andarthrodires, which went extinct at that time,were primarily shearing predators; the Mis-sissippian chondrichthyans and actinoptery-gians that replaced them were generallycrushing predators (Sallan and Coates 2010).Corroborating this, angular shell fragments ofthe type produced by crushing predationbecame more common after the Hangenbergextinction (Salamon et al. 2014). Sallan et al.(2011) additionally conclude that the Tournai-sian–Visean peak in crinoid diversity was areaction among camerates to the disappear-ance of Devonian predatory fish during theend-Devonian Hangenberg extinction.

If the dominant mode of predation changedfrom nonlethal grazing to crushing at thistime, then the higher EAL in cameratesoriginating in the Tournaisian may have beena response to the relaxation of selectivepressure from that form of predation. Anti-predatory crinoid arm morphologies are likelyto have been less effective against crushingpredators, and as these predators becamedominant after the Devonian, crinoids mighthave responded to them instead; a peak incamerate spinosity in the Mississippian (Si-gnor and Brett 1984) and a driven trend inmonobathrid camerates toward a decreasingvariety and number of ossicles in the calyxduring the end-Devonian extinction (Simpson2010) can both be interpreted as specificallyanti-crushing defenses.

Conclusions

Arm morphologies well adapted to surviv-ing frequent arm loss became increasinglycommon in camerate crinoids during thePaleozoic. However, cladids, the second larg-est taxon, did not exhibit a similar trend. Aplausible explanation for these contrastingpatterns is that predation leading to arm losswas greater on camerates than cladids, con-sistent with evidence that the frequency ofarm loss and regeneration both was higher incamerates and increased in camerates duringthis period (Baumiller and Gahn 2004; Gahnand Baumiller 2010).

A possible reason for the taxonomic differ-ence in adaptation to predation is the ob-served preference of platyceratid snails forcamerates, and for particular taxa of camerates(Ausich 1980; Baumiller and Gahn 2004; Gahnand Baumiller 2006). If these gastropods, andperhaps other crinoid infesters, were theprimary targets of predators (Meyer 1985;Brett 2003; Hempson and Griffiths 2008), itcould incur incidental damage to their hostsand provide selective pressure toward preda-tor-resistant arm morphologies.

The camerate EAL values plateau by theMiddle Devonian, corresponding possibly to anatural minimum in the adaptive value of armpatterns and possibly to an ecological shift inpredator strategy. A shift in predatory strate-gies on crinoids is likely given the post-Devonian change in dominance of predatoryfishes that made anti-grazing adaptations lesseffective and instead favored anti-crushingadaptations (Signor and Brett 1984; Watersand Maples 1991; Sallan and Coates 2010;Simpson 2010; Sallan et al. 2011).

Acknowledgments

This work was partially funded by grantsfrom the National Science Foundation (DEB1036393; EAR 0824793) and the NationalGeographic Society (NGS 8505-08). The au-thors thank J. M. Koniecki (www.crinus.info)for kindly providing access to his extensivefossil crinoid collection; reviewers D. Meyerand T. Oji for their insightful comments andsuggestions; K. J. Rhodes and E. M. Moacdiehfor helpful discussions during preparation;and the Case Award committee at the Uni-versity of Michigan for reviewing an earlyversion of this manuscript. Many of the ideasaddressed in this paper were influenced bydiscussions with F. J. Gahn, who also providedthe photos in Figure 1.

Literature Cited

Ausich, W. I. 1980. A model for niche differentiation in Lower

Mississippian crinoid communities. Journal of Paleontology

54:273–288.

Ausich, W. I., and E. W. Roeser. 2012. Camerate and disparid

crinoids from the Late Kinderhookian Meadville Shale, Cuya-

hoga Formation of Ohio. Journal of Paleontology 86:488–507.

Bambach, R. K. 1999. Energetics in the global marine fauna: a

connection between terrestrial diversification and change in the

marine biosphere. Geobios 32:131–144.

426 V. J. SYVERSON AND TOMASZ K. BAUMILLER

Baumiller, T. K. 1993. Survivorship analysis of Paleozoic Crin-

oidea: effect of filter morphology on evolutionary rates.

Paleobiology 19:304–321.

———. 2008. Crinoid ecological morphology. Annual Review of

Earth and Planetary Sciences 36:221–249.

———. 2013. Ephemeral injuries, regeneration frequencies, and

intensity of the injury-producing process. Marine Biology

160:3233–3239.

Baumiller, T. K., and F. J. Gahn. 2004. Testing predator-driven

evolution with Paleozoic crinoid arm regeneration. Science

305:1453–1455.

Baumiller, T. K., R. Mooi, and C. G. Messing. 2008. Urchins in the

meadow: paleobiological and evolutionary implications of

cidaroid predation on crinoids. Paleobiology 34:22–34.

Bottjer, D. J., and D. Jablonski. 1988. Paleoenvironmental patterns

in the evolution of post-Paleozoic benthic marine invertebrates.

Palaios 3:540–560.

Brett, C. E. 2003. Durophagous predation in Paleozoic marine

benthic assemblages. Pp. 401–432 in P. H. Kelley, M. Kowalew-

ski, and T. A. Hansen, eds. Predator-prey interactions in the

fossil record. Springer, New York.

Brett, C. E., and S. E. Walker. 2002. Predators and predation in

Paleozoic marine environments. In M. Kowalewski and P. H.

Kelley, eds. The fossil record of predation. Paleontological

Society Special Papers 8:93–118.

Brett, C. E., F. J. Gahn, and T. K. Baumiller. 2004. Platyceratid

gastropods as parasites, predators, and prey and their possible

effects on echinoderm hosts: collateral damage and targeting.

Geological Society of America Abstracts with Programs 36:478.

Brower, J. C. 2006. Ontogeny of the food-gathering system in

Ordovician crinoids. Journal of Paleontology 80:430–446.

Cowen, R. 1981. Crinoid arms and banana plantations: an

economic harvesting analogy. Paleobiology 7:332–343.

Dahl, T. 2010. Devonian rise in atmospheric oxygen correlated to

the radiations of terrestrial plants and large predatory fish.

Proceedings of the National Academy of Sciences USA

107:17,911–17,915.

Gahn, F. J., and T. K. Baumiller. 2005. Arm regeneration in

Mississippian crinoids: evidence of intense predation pressure

in the Paleozoic? Paleobiology 31:151–164.

———. 2006. Using platyceratid gastropod behaviour to test

functional morphology. Historical Biology 18:397–404.

———. 2010. Evolutionary history of regeneration in crinoids

(Echinodermata). Integrative and Comparative Biology 50:514a–

514m.

Gorzelak, P. L., L. Rakowicz, M. A. Salamon, and P. Szrek. 2011.

Inferred placoderm bite marks on Devonian crinoids from

Poland. Neues Jahrbuch fur Geologie und Palaontologie,

Abhandlungen 259:105–112.

Hempson, T. N., and C. L. Griffiths. 2008. Symbionts of comatulid

crinoids in False Bay, South Africa. African Zoology 43:237–244.

Janevski, G. A., and T. K. Baumiller. 2010. Could a stalked crinoid

swim? a biomechanical model and characteristics of swimming

crinoids. Palaios 25:588–596.

Kammer, T. W., and W. I. Ausich. 1987. Aerosol suspension

feeding and current velocities: distributional controls for Late

Osagean crinoids. Paleobiology 13:379–395.

Lane, N. G. 1984. Predation and survival among inadunate

crinoids. Paleobiology 10:453–458.

Lawrence, J. M. 2009. Arm loss and regeneration in stellate

echinoderms: an organismal view. Pp. 53–66 in C. Johnson, ed.

Echinoderms in a changing world. Proceedings of the 13th

International Echinoderm Conference. Hobart, Tasmania, Aus-

tralia.

Lawrence, J. M., and J. Vasquez. 1996. The effect of sublethal

predation on the biology of echinoderms. Oceanologica Acta

19:431–440.

McClintock, J. B., B. J. Baker, T. K. Baumiller, and C. G. Messing.

1999. Lack of chemical defense in two species of stalked

crinoids: support for the predation hypothesis for Mesozoic

bathymetric restriction. Journal of Experimental Marine Biology

and Ecology 232:1–7.

Meyer, D. L. 1985. Evolutionary implications of predation on

recent comatulid crinoids from the Great Barrier Reef. Paleobi-

ology 11:154–164.

Meyer, D. L., and D. B. Macurda. 1977. Adaptive radiation of the

comatulid crinoids. Paleobiology 3:74–82.

Meyer, D. L., C. A. LaHaye, N. D. Holland, A. C. Arneson, and J.

R. Strickler. 1984. Time-lapse cinematography of feather stars

(Echinodermata: Crinoidea) on the Great Barrier Reef, Australia:

demonstrations of posture changes, locomotion, spawning and

possible predation by fish. Marine Biology 78:179–184.

Mladenov, P. V. 1983. Rate of arm regeneration and potential

causes of arm loss in the feather star Florometra serratissima

(Echinodermata: Crinoidea). Canadian Journal of Zoology

61:2873–2879.

Moore, R., H. W. Rasmussen, N. G. Lane, G. Ubaghs, H. L.

Strimple, R. E. Peck, J. Sprinkle, R. O. Fay, and H. Sieverts-

Doreck. 1978. Systematic Descriptions. Pp.T402–T812 in G.

Ubaghs et al. Echinodermata 2, Crinoidea. Part T of R. C. Moore

and C. Teichert, eds. Treatise on invertebrate paleontology.

Geological Society of America, Boulder, Colo. and University of

Kansas, Lawrence.

Nichols, D. 1994. Reproductive seasonality in the comatulid

crinoid Antedon bifida (Pennant) from the English Channel.

Philosophical Transactions of the Royal Society of London B

343:113–134.

Oji, T. 2001. Fossil record of echinoderm regeneration with special

regard to crinoids. Microscopy Research and Technique 55:397–

402.

Oji, T., and T. Okamoto. 1994. Arm autotomy and arm branching

pattern as anti-predatory adaptations in stalked and stalkless

crinoids. Paleobiology 20:27–39.

Rideout, J. A., N. B. Smith, and M. D. Sutherland. 1979. Chemical

defense of crinoids by polyketide sulphates. Experientia

35:1273–1274.

Salamon, M. A., P. Gorzelak, R. Niedzwiedzki, D. Trzesiok, and T.

K. Baumiller. 2014. Trends in shell fragmentation as evidence of

mid-Paleozoic changes in marine predation. Paleobiology 40:1–10.

Sallan, L. C., and M. I. Coates. 2010. End-Devonian extinction and

a bottleneck in the early evolution of modern jawed vertebrates.

Proceedings of the National Academy of Sciences USA

107:10,131–10,135.

Sallan, L. C., T. W. Kammer, W. I. Ausich, and L. A. Cook. 2011.

Persistent predator-prey dynamics revealed by mass extinction.

Proceedings of the National Academy of Sciences USA

108:8335–8338.

Schneider, J. A. 1988. Frequency of arm regeneration of comatulid

crinoids in relation to life habit. Pp. 531–538 in B. F. Keegan and B.

D. S. O’Connor, eds. Echinoderm biology. Proceedings of the Sixth

International Echinoderm Conference. Balkema, Rotterdam.

Signor, P. W., and C. E. Brett. 1984. The mid-Paleozoic precursor to

the Mesozoic Marine Revolution. Paleobiology 10:229–245.

Simpson, C. 2010. Species selection and driven mechanisms jointly

generate a large-scale morphological trend in monobathrid

crinoids. Paleobiology 36:481–496.

Vermeij, G. J. 1977. The Mesozoic marine revolution: evidence

from snails, predators and grazers. Paleobiology 3:245–258.

Waters, J. A., and C. G. Maples. 1991. Mississippian pelmatozoan

community reorganization: a predation-mediated faunal

change. Paleobiology 17:400–410.

Webster, G. D. 2003. Bibliography and index of Paleozoic crinoids,

coronates, and hemistreptocrinoids, 1758–1999. Geological

Society of America Special Paper 363.

IMPACT OF PREDATION ON CRINOID ARMS 427