the stability of thoracic segmentation in trilobites: a ... · hence trilobites from the entire...

EVOLUTION & DEVELOPMENT

1:1, 24–35 (1999)

©

BLACKWELL SCIENCE, INC.

24

The stability of thoracic segmentation in trilobites: a case study in

developmental and ecological constraints

Nigel C. Hughes,

a,

* Ralph E. Chapman,

b

and Jonathan M. Adrain

c

a

Department of Earth Sciences, University of California, Riverside, California 92521 USA;

b

ADP-136, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560 USA;

c

Department of Geology, University of Iowa, Iowa City, Iowa 52242 USA

*Author for correspondence (email: nigel.hughes@ucr.edu)

SUMMARY

The decline in origination rate of new meta-zoan body plans following the Cambrian radiation has beensuggested to reflect developmental canalization in derivedtaxa, limiting their ability to evolve forms with radically differ-ent morphotypes. Segmentation is a fundamental aspect ofarthropod body plan, and here we show that a derived trilo-bite that secondarily converged on a morphotype character-istic of basal members of the clade also reverted to a pattern

of segmental variability common among basal trilobites.Hence a secular trend in loss of variability of the trilobite tho-rax was not due to the evolution of an inviolable developmentalconstraint. This result challenges the notion of developmen-tal canalization in phylogenetically derived taxa. Rather, earlyvariability in trilobites may be the result of ecological factorsthat promoted segment-rich thoracic morphotypes duringCambrian time.

INTRODUCTION

Specification of segmentation occurs early in the hierarchyof genetic controls that guide individual development (seeRaff 1996; Sommer and Tautz 1993), and many aspects ofthis process are conserved throughout the Arthropoda(Akam et al. 1994; Carroll 1995). Details of the form and de-velopment of segments are preserved in a variety of fossil ar-thropods and offer the potential for insight into the evolu-tionary history of developmental systems (Arthur 1997; Raff1996). Here we use the fossil record of an outstandingly wellpreserved trilobite assemblage to investigate the controls ofa major clade-wide reduction in the variability of the numberof segments in the thorax. Specifically, we test one aspect ofthe hypothesis that developmental patterns became cana-lized or “hardened” after the early Cambrian, preventing fur-ther rapid diversification (McNamara 1983, 1986; McKin-ney and McNamara 1991). More generally, variation in theconstraint of segment numbers is widely reported within ex-tant groups, including crustaceans, centipedes, annelids, andvertebrates (e.g., Bateson, 1894; Minelli and Bortoletto1988; Sassaman et al. 1997; Schram 1986), and may providea model system for investigating the evolution of develop-mental characters governing bodyplan design (Arthur 1999).

TEMPORAL TRENDS IN THE DEVELOPMENT OF THE TRILOBITE THORAX

Trilobites were extant from the early Cambrian (about 525million years ago) until the end of the Permian (about 250million years ago). Many thousands of species have been de-scribed, and hundreds of these are known from completelyarticulated dorsal exoskeletons. Several Cambrian species atbasal positions within the clade show variable numbers ofthoracic segments, with adults of the same species differingby two or more segments in the thorax (McNamara 1983;Stubblefield 1959) (Table 1). Early species also differgreatly in the total number of thoracic segments, with Cam-brian forms ranging between two and more than 40 thoracicsegments in adulthood. This pattern contrasts markedly withthat seen in later, phylogenetically derived, trilobite groupsin which the number of adult thoracic segments range fromsix to 22 segments and are constant not only within species,but often at the family level or higher (Hughes and Chapman1995; McNamara 1983; Stubblefield 1959). Hence variationin thoracic segment numbers apparently became progres-sively constrained during trilobite evolution, both at the levelof individual species and at higher taxonomic ranks. Further-more, it has been suggested that the control of thoracic seg-

Hughes et al.

Trilobite segmentation constraint

25

mentation became irreversibly “hardened” or entrenched inderived taxa (McNamara 1983; McKinney and McNamara1991) limiting their opportunity to evolve morphotypes withradically different body plans. If irreversible entrenchmentduring trilobite evolution could be demonstrated this wouldprovide support for the notion of early plasticity in primitivegroups followed by later, advanced, constraint.

THE LOD NICE TRILOBITE ASSEMBLAGE

To test this idea we have examined the relationship betweenphylogenetic ancestry and segmentation pattern in an excep-tionally fossiliferous assemblage of Silurian trilobite speciesbelonging to a variety of clades (Figs. 1, 2.1–2.3, and 5B).This assemblage is about 425 million years old and post-dates the end of the Cambrian by approximately 60 millionyears. The particular taxonomic and preservational charac-teristics of this assemblage make it an excellent candidate fordissecting the detailed patterns of variation among contem-porary individuals: a relatively rare opportunity in the fossilrecord. Each Lod nice species examined belonged to a cladewhose basal members had stable numbers of thoracic seg-ments in adulthood. This assemblage contained nine trilobitespecies represented by articulated exoskeletons (K í 1992).These were collected from numerous individual beddingplanes within a 1.4-meter statigraphic interval at a localitynear Lod nice in the Czech Republic. Because the speci-mens came from several surfaces separated by interveningsediment, all were not living at the same instant and the sam-

E

e

r z

e

ple is “time-averaged.” Given this, it is important to assess:(1) how much time did it take for trilobites to accumulate onindividual bedding planes; (2) over what period did the 1.4-meter interval accumulate; and (3) do trilobites collectedfrom different bedding surfaces show different patterns ofvariation.

Taphonomic evidence on the preservational condition ofthe specimens suggests that all well-preserved individualswere autochthonous to the area and accumulated on individ-ual bedding surfaces in periods of days to months (Hughesand Chapman 1995). Based on the depositional fabric andaverage accumulation rates for comparable environments,the entire interval accumulated in a series of approximately200–400 short-lived depositional events, probably the resultof individual storms, spread over a period of between 1,000to 10,000 years (Sadler and Strauss 1990). Measured collec-tions came from several bedding planes but all taxa likely ex-perienced a similar degree of time-averaging and compaction-related deformation. The majority of specimens analyzedherein were collected by Joachim Barrande in the 1840s, andtheir exact position within the 1.4-meter interval cannot beestablished with confidence. Nevertheless, analyses of mul-tiple specimens of

Aulacopleura konincki

on single slabs,and of new collections made from in situ bedding planes,suggest that the degree of variation in segmentation patternand overall shape is similar both within and between collec-tions from different bedding plane surfaces. Hence trilobitesfrom the entire interval could be treated as one sample. Fur-thermore, regardless of the degree of time averaging, all spe-cies within the assemblage shared a similar accumulation

Table 1. Intraspecific variation in numbers of thoracic segments in adult trilobites in the relatively rare species that exhibit variation in this character (Note that the great majority of examples are of Cambrian age)

Order/Suborder Family Species Age

Range insegmentnumbers Range

Meansegmentnumber Reference

Redlichiida Olenellidae

Olenellus thompsoni

Early Cambrian 18–19 2 18.5 Whittington 1989Redlichiida Paradoxididae

Paradoxides davidis

Middle Cambrian 18–21 4 19.5 Bergström and Levi-Setti 1978Redlichiida Ellipsocephalidae

Ellipsocephalus polytomus

Middle Cambrian 13–14 2 13.5 Westergård 1936Agnostida Eodiscidae

Pagetia triaena

Middle Cambrian 2–3 2 2.5 Jell 1975Agnostida Eodiscidae

Pagetia polygnota

Middle Cambrian 2–3 2 2.5 Jell 1975“Ptychopariida” Ptychoparidae

Ptychoparia anderseni

Middle Cambrian 12–13 2 12.5 Henningsmoen 1952“Ptychopariida” Alokistocaridae

Elrathia kingi

Middle Cambrian 10–13 4 11.5 Bright 1959“Ptychopariida” Alokistocaridae

Elrathia antiquata

Middle Cambrian 13–14 2 13.5 Schwimmer 1989“Ptychopariida” Alokistocaridae

Alokistocare idahoense

Middle Cambrian 22–24 3 23 Gunther and Gunther 1981“Ptychopariida” Alokistocaridae

Alokistocare laticaudum

Middle Cambrian 17–18 2 17.5 Gunther and Gunther 1981“Ptychopariida” Annamitiidae

Jenkinsonia varga

Middle Cambrian 12–14 3 13 Robison 1971“Ptychopariida” Asphelaspidae

Labiostria westropi

Upper Cambrian 12–13 2 12.5 Chatterton and Ludvigsen 1998“Ptychopariida” Pterocephaliidae

Pterocephalia norfordi

Upper Cambrian 12–13 2 12.5 Chatterton and Ludvigsen 1998Olenina Olenidae

Wujiajiania sutherlandi

Upper Cambrian 13–14 2 13.5 Chatterton and Ludvigsen 1998Olenina Olenidae

Eucarye latum

Upper Cambrian 16–17 2 16.5 Henningsmoen 1957Olenina Olenidae

Olenus cataractes

Upper Cambrian 14–15 2 14.5 Hughes and Chapman 1995Olenina Olenidae

Olenus attenuatus

Upper Cambrian 15–16 2 15.5 Hughes and Chapman 1995Olenina Olenidae

Triarthrus eatoni

Upper Ordovician 13–14 2 13.5 Cisne 1981Proetida Aulacopleuridae

Aulacopleura konincki

Middle Sliurian 18–22 5 20 Hughes and Chapman 1995Proetida Aulacopleuridae “

Otarion diffractum

” Middle Silurian 17–20 4 18.5 This article

26 EVOLUTION & DEVELOPMENT

Vol. 1, No. 1, July–Aug. 1999

Fig. 1. Adults of trilobite speciesfrom the Lod nice assemblage an-alyzed for intraspecific variation.(1a,b) Cheirurus insignis [OrderPhacopida] a. Small adult with 11thoracic segments, NMPL15749; b.Large adult with 11 thoracic seg-ments, MCZ7504/6a. (2) Theban-aspis putzkeri [Order Proetida]with 10 thoracic segments, NMPL

E257. (3a–c) Aulacopleura kon-incki [Order Proetida]. a. Smalladult with 19 thoracic segments,MCZ103490; b. Large adult with20 thoracic segments, and posi-tions of landmarks used in shapeanalysis outlined, BMNH42367.4;c. Intermediate-sized adult with 22thoracic segments MCZ103494.(4) Odontopleura ovata [OrderOdontopleurida] with nine thorac-ic segments NMPL E260. (5a,b)Planiscutellum planum [Order Cory-nexochida]. a. Small adult with 10thoracic segments MCZ3871; b.Large adult with 10 thoracic seg-ments NMPL E265. (6) Scharyiawenlockiana [Order Proetida] with6 thoracic segments MCZ5311.White arrows indicates position ofthoracic/pygidial boundary. Scalebar equals 2 mm in all specimens.Repository information: BNMH 5Natural History Museum, Lon-don; MCZ 5 Museum of Compar-ative Zoology, Harvard University;NMPL 5 National Museum, Pra-gue; CGU 5 Czech GeologicalSurvey.

e

C

C

C

and preservational history, and thus can be readily com-pared. Six species were preserved in sufficient numbers formetric comparison of growth patterns.

Geometric relationships between a set of 22 morphologi-cal landmarks (Fig. 1.3b), common to all taxa and distributedthroughout the dorsal exoskeleton, were examined to assessthe degree of morphological variation in shape amongadults. This analysis confirmed that variation among the in-dividuals assigned to each species was continuous, with noevidence of polymorphism. The degree of intraspecific vari-

ation within adults of each taxon was measured as the vari-ance of individuals in the population about the mean form forthat species (Fig. 3A). All taxa show comparable degrees ofmorphologic variation, and the least variable species isthe proetide

Aulacopleura konincki

(Fig. 1.3a–c). A resam-pling procedure, used to examine the relationship betweenobserved morphological range and sample size, confirmedthat differences in variability were not the product of sam-ple-size differences among species (Fig. 3B). If growth wassignificantly allometric, morphologic diversity within spe-

Hughes et al.

Trilobite segmentation constraint

27

cies could be affected by the range of adult sizes. This is notthe case in this analysis. In Fig. 3A the range of adult size in-creases toward the bottom of the chart, but degree of shapevariation does not reflect this order. (Furthermore, the slopesof all pairwise comparisons of specimens in plots of shapedifference against difference in centroid size never exceed0.006, implying virtual isometry.) These results indicatethat, in terms of overall shape, each species shows a compa-rable degree of variation. Because there is minimal variationin meristic characters within any of the species (except thatdescribed below in

A. konincki

) each Lod nice species likelyrepresents a single biological species, exhibiting continuousvariation among constituent individuals.

The post-cephalic region of trilobites grew by generatingnew segments from a terminal teloblast in the pygidium, andby increasing the size of previously formed segments. Theonset of adulthood is defined by a sharp decline in the rate ofsegment addition (Hughes and Chapman 1995), and all trilo-

e

bites continued to molt and grow larger during adulthood.Each Lod nice species followed a segment accumulationtrajectory prior to the onset of adulthood, with a strong cor-respondence between number of segments and glabellarlength (Fig. 4). Glabellar length is a suitable proxy for over-all size because cephalic segmentation is constant during theperiod of ontogeny considered herein and these segments,once formed, grew isometrically. [Analysis of the metric di-mensions of

A. konincki

indicates that glabellar length showsthe highest loadings on an isometric principal component 1(Hughes and Chapman 1995), confirming the utility of thismeasure]. All species with less than 14 segments in adult-hood had a constant number of segments after attaining theadult growth stage (Fig. 4), but one species,

Aulacopleurakonincki

, possessed an unusually extended pre-adult ontog-eny with an anomalously large number of thoracic segmentsfor any post-Cambrian trilobite. This species also showsmarked variation in segment numbers in adulthood, with be-

e

Fig. 2. 1–3 Additional adult trilo-bites from Lod nice. (1) Phaco-pidella glocerki [Order Phacopida]adult with 11 thoracic segments,NMPL17005. (2) Dicranopeltis sca-bra [Order Lichida] with 11 tho-racic segments (3) Decoroproetuscoderus [Order Proetida] with 10thoracic segments. 4–6. Adults ofCambrian olenimorphic trilobiteshomeomorphic with Aulacopleurakonincki. (4) Elrathia kingi [Order“Ptychopariida”] with 13 thoracicsegments, Middle Cambrian, Utah,USNM476861. (5) Alokistocare ida-hoense [Order “Ptychopariida”]with 22 thoracic segments, MiddleCambrian, Utah, Gunter privatecollection. (6) Ptychoparia striata[Order “Ptychopariida”] with 14thoracic segments, Middle Cam-brian, Jince, Czech Republic, CGUM 328. Inverted commas impliesparaphyletic group. White arrowindicates position of thoracic/py-gidial boundary. Scale bar equals 2mm in all specimens.

e

S

28 EVOLUTION & DEVELOPMENT

Vol. 1, No. 1, July–Aug. 1999

tween 18 to 22 thoracic segments. It is the only trilobiteknown to increase the number of thoracic segments through-out the adult growth stage (Hughes and Chapman 1995), al-though markedly more slowly than the rate during pre-adultgrowth, assuming that glabellar length change was approxi-

mately constant with time. This trend, and the fact that thisspecies shows the lowest degree of variation in the geometricanalysis, strongly suggest that all specimens are members ofa single species showing continuous variation among char-acter states.

Fig. 3. (A) All species show comparable levels of intraspecific variation as indicated by the spread of shape differences of individuals frommean geometry for each Lod nice species. Comparison based on 22 homologous landmark positions on the dorsal exoskeleton of sixtrilobite species using least squares Procrustes fits (LSTRA) (Chapman 1990). Horizontal bars are standard deviations of specimen dis-tribution about the mean distance. Numbers of specimens analyzed for each species are to left of standard deviation bars. (B) Relation-ship between sample size and degree of intraspecific variation in three Lod nice taxa, showing that the degree of shape variation is largelyindependent of sample size. At each sample increment of five additional specimens 20 samples were drawn randomly and for each ofthese the range of their average shape distance from the mean was calculated. Vertical bars express the standard deviation of the distri-bution of these 20 ranges for each sample increment. Generic abbreviations Al. 5 Alokistocare, Pt. 5 Ptychoparia, El. 5 Elrathia, Ph. 5Phacopidella, C. 5 Cheirurus, S. 5 Scharyia, P. 5 Planiscutellum, T. 5 Thebanaspis, D. 5 Decoroproetus, Di. 5 Dicranopeltis, O. 5 Od-ontopleura. Specimens used in these analyses included museum collections, notably those of J. Barrande, and those collected in the fieldby N. C. H. and Ji í K í . The Lod nice assemblage includes a number of other articulated trilobite species but these are not sufficientlycommon or well-preserved to permit morphometric analysis.

e

e

r r z e

Hughes et al.

Trilobite segmentation constraint

29

That

A. konincki

alone among these species shows an un-usual pattern of thoracic segmentation points to a linkage be-tween its uniquely segment-rich morphotype and thoracicvariability. Whatever promoted variation in segment number,be it environmental factors or otherwise, was restricted in itseffect to this taxon among the Lod nice fauna. Hence this as-semblage provides an excellent example of differential pat-terns of intraspecific variation among trilobite taxa, with anunusually tight control for the environmental setting. The ob-jective now becomes to understand the cause and significanceof the unique pattern of variation in

A. konincki.

MORPHOTYPIC AND ECOLOGIC ASPECTS OF

A. KONINCKI

In overall morphology

A. konincki

is characterized by a nar-row axis, extended and caecate frontal area and pleurae, andsegment-rich thorax. This form is referred to as the “oleni-morphic” morphotype and its origin has been related to con-ditions of reduced oxygen availability (Fortey and Owens1990). However, because

A. konincki

co-occurs at Lod nicewith a range of other trilobite morphotypes, in addition toother fauna indicative of normal benthic conditions, it is un-likely that dysaerobic conditions persisted at the time of dep-osition of these sediments. Nevertheless, the specific nicheoccupied by

A. konincki

may have been related to lower lev-els of available oxygen. Possibilities for such a condition in-clude an infaunal burrowing habit into poorly oxygenatedsubstrate (although the preservation of undisturbed fine-

e

e

scale sediment lamination makes this unlikely), or temporaryperiods of oxygen deficiency, during which

A. konincki

flourished to the exclusion of other taxa. Regardless of itsspecific niche, the structure of the mouth plate suggests that

A. konincki

fed by the processing of fine particulate material,similar to that invoked for all olenimorphs, and many addi-tional primitive libristomate trilobites (Fortey 1990). Otheroccurrences of Silurian

Aulacopleura

are geographicallywidespread, and generally found in low diversity assem-blages deposited in deeper water, basinal facies (Thomas andLane 1984). For example, an

Aulacopleura-Raphiophorus

-odontopleuride association occurs in early Silurian rocksfrom New South Wales, Australia (G. D. Edgecombe, per-sonal communication, 1998), Bathurst Island in the Cana-dian Arctic, and South China (Wang 1989). As today, theselocalities were widely dispersed during Silurian time.

The striking morphological similarity of

A. konincki

withCambrian olenimorphs has long been noted (Fortey 1990;P ibyl 1947) and offers the opportunity to observe patterns ofvariation in a common Cambrian morphotype alive long afterthe close of Cambrian time. Phenetic comparisons (Figs. 5Aand 6) of

A. konincki

with other members of the Lod nice tri-lobite assemblage and selected Cambrian olenimorphic trilo-bites (Figs. 2.4–2.6) confirm that

A. konincki

groups witholenimorphic species. Indeed, in the cluster analysis (Fig.5A) the similarity of

A. konincki

to each and every selectedCambrian olenimorphic species is greater than the similarityfound in any pairwise comparison of Lod nice trilobite spe-cies. Non-metric multidimensional scaling and nearest neigh-bor analysis (Fig. 6) confirm that

A. konincki

occupies the

r

e

e

Fig. 4. Relationship of number ofthoracic segments to overall size,as represented by the length of theglabella. After the onset of adult-hood (recognized as a sharp changein segment accretion per unit ceph-alic growth) all taxa, except Aula-copleura konincki, had constantnumbers of segments regardless ofoverall size. In marked contrast A.konincki shows 18–22 segments,and a significant tendency to in-crease segment number throughoutadult growth (p , 0.01) (Hughesand Chapman 1995). Generic ab-breviations are as in the legend toFig. 3.

30 EVOLUTION & DEVELOPMENT

Vol. 1, No. 1, July–Aug. 1999

area of morphospace adjacent to the Cambrian olenimorphs.Hence the grouping is not the result of distortions of pheneticspace that can be induced by clustering metrics.

The convergent form of

A. konincki

and Cambrian oleni-morphs raises two questions. The first is whether

A. konincki

and the Cambrian olenimorphs share a common form be-cause they share a close phylogenetic origin, or whether thissimilarity is secondarily convergent. The second question ishow patterns of intraspecific variation in thoracic segment

numbers map onto the phylogenetic relationships among theLod nice and other taxa.

PHYLOGENETIC PLACEMENT OF A. KONINCKI

High-level classification of trilobites is relatively poorly re-solved. Although recent work (Fortey 1990, 1997; Forteyand Chatterton 1988) has now accommodated most families

e

Fig. 5. Contrasting phenetic similarity ofoverall form in the study sample and phy-logenetic relationships, showing the con-vergence of A. konincki on a primitivemorphotype. Lod nice taxa are givenshaded (Order Proetida) or solid bars(non-proetides); Cambrian homeomorphsof A. konincki taxa have open bars. Num-bers of thoracic segments are indicated be-neath bars. (A) Morphologic similarity ofA. konincki and Cambrian olenimorphicspecies based on UPGMA cluster analysisof the distance matrix derived from theleast squares Procrustes fit of 24 homolo-gous landmarks (Chapman 1990). A. koninckispecimens and the Cambrian olenimorphsshare greater phenetic similarity than anypairwise comparison among Lod nicetaxa. The phenetic analysis links othermembers of the proetide clade (shadedbars). A. konincki1 is a large adult, A.konincki2 is the average holaspid form forthis species (Hughes and Chapman 1995).(B) Trilobite phylogenetic relationshipsindicate that A. konincki is a derived trilo-bite, with close relatives characterized bysmall numbers of the thoracic segments.The topology is based on current views oftrilobite phylogenetics (Fortey 1990, 1997)and developmental information (Chatter-ton and Speyer 1997). Node 1 is character-ized by adult-like larvae with prominent,forward-expanding glabellae (Order Cory-nexochida). Node 2 is characterized byboxy, deep larvae with a corona of bifurcatemarginal spines, longitudinally subdividedglabellar furrows, and paired cephalic tu-bercles (Order Phacopida). Node 3 is char-acterized by the ptychoparioid larval type(Order “Ptychopariida”). Node 4 is charac-terized by a larva bearing a preglabellarfield (Order Proetida). Node 5 is charac-terized by the possession of only two larvalstages, with metamorphosis between them(Edgecombe et al. 1997). Node 6 is charac-terized by possession of an aulacopleuroidlarvae, typically with four protaspid stagesand a distinctive anaprotaspid. Ordinallevel names given along branches, “Pty-chopariida” is paraphyletic. Generic ab-breviations are as in the legend to Fig. 3.

e

e

Hughes et al. Trilobite segmentation constraint 31

in ordinal groups, the majority of Cambrian orders are para-phyletic and the phylogenetic relationships between Cam-brian and post-Cambrian orders remain largely obscure.Taken together, the Lod nice taxa form a representativecross section of post-Cambrian trilobite diversity. Althoughthe general lack of phylogenetic resolution reflects the scaleof the remaining high-level problems, those relationshipswhich are depicted in Fig. 5B are very robustly supportedand generally accepted. Intriguingly, A. konincki is not thesister taxon of the Cambrian olenimorphs, but lies nestedwithin a group of phylogenetically derived trilobites. TheLod nice assemblage contains four members of this clade,which together represent the Order Proetida. The three otherproetide species cluster together in phenetic space (Figs. 5Aand 6) and are here referred to as the non-olenimorphic pro-etides. The fact that A. konincki has nearest relatives at Lod -nice that are non-olenimorphic is strong evidence that the re-

e

e

e

semblance between the Silurian and Cambrian olenimorphsis the product of secondary convergence. However, the truedegree of phylogenetic separation between these olenimor-phic groups is best appreciated in the context of proetide in-group relationships (Fig. 7, an expansion of node 3 in Fig. 5B).

Although proposed nearly a quarter century ago (Forteyand Owens 1975), the most compelling argument for themonophyly of Proetida (and therefore the phylogenetic sep-aration of A. konincki from the Cambrian olenimorphs) isFortey’s (1990) observation that the group is characterizedby the developmental displacement of the natant hypostomalcondition (an adult character of the broad libristomate clade)and its fixation in earliest ontogenetic stages. Proetide larvaeare unique among trilobites in the possession of a preglabel-lar field, a reflection of their free ventral mouth plate. Whilethe composition and validity of Proetida is thus established,there have been no comprehensive analyses of ingroup phy-logeny. The broader phylogenetic context of Lod nice proe-tide trilobites and the Cambrian olenimorphs (Fig. 7), illus-trates the considerable phylogenetic distance between A.konincki and its Cambrian homeomorphs. Fig. 7 representsthe preliminary result of work in progress on this problem,based on recent field discoveries and cladistic analyses byJ. M. A. [see also Adrain and Chatterton (1993)].

Two points bear emphasis. First, Aulacopleura is phylo-genetically extremely distant from all known Cambrian ole-nimorph taxa. Basal taxa of Proetida have recently beenidentified through new ontogenetic evidence and these, themarjumioideans, are of late Middle Cambrian age. Thismeans that there is strong evidence that Aulacopleura canonly share very ancient, at the youngest early Middle Cam-brian, common ancestry with the olenimorphs. It is separatedfrom this presumptive ancestry by a plethora of major phy-logenetic steps comprising the genesis of several superfamil-ial clades. The second point is that none of these other proet-ide taxa are olenimorphs. In fact, with the exception of somederived members of the Otarioninae, sister group of Aulaco-pleurinae, none of these major groups includes species withgreater than 12 thoracic segments, and most clades (includ-ing the phylogenetically basal Proetoidea) have 10 or less.Proetides, in general, have convex, arched exoskeletons, of-ten with a coarsely tuberculate sculpture, relatively narrowpleurae, and modest of low segment counts.

Hence, there is little doubt that Aulacopleura belongs to ahighly derived trilobite clade that became secondarilyadapted to the ecological niche exploited by the olenimor-phs. This niche was continuously occupied during the Or-dovician by other major trilobite clades (including olenidsthemselves and asaphids). The elimination of these groupsduring the great end-Ordovician mass extinction may havepaved the way for a surviving group to exploit dysaerobicenvironments, and to secondarily acquire olenimorphic mor-phology.

e

Fig. 6. Clustering of Aulacopleura konincki with Cambrian oleni-morphs, and significant separation from non-olenimorphic proe-tides in a non-metric multidimensional scaling. Analysis based onthe least squares Procrustes distance matrix of the specimens an-alyzed in Fig. 5A. Proportion of total variation represented by thistwo-dimensional model: 92.6%, stress: 0.117. Note that, as in Fig.5A, all olenimorphs plot closer to the phacopids than to the non-olenimorphic proetides. Nearest-neighbor analysis confirms thatthe two specimens of A. konincki are nearest neighbors. Specimen1 is the nearest neighbor of Al. idahoense, and specimen 2 is thenearest neighbor of E. kingi. Abbreviations: A 5 Aulacopleurakonincki, Al 5 Alokistocare idahoense, Pt 5 Ptychoparia striata,E 5 Elrathia kingi, C 5 Cheirurus insignis, g 5 Phacopidella glo-cerki, T 5 Thebanaspis putzkeri, D 5 Decoroproetus coderus, w 5Scharyia wenlockiana, s 5 Dicranopeltis scabra, O 5 Odonto-pleura ovata, p 5 Planiscutellum planum.

32 EVOLUTION & DEVELOPMENT Vol. 1, No. 1, July–Aug. 1999

INTERPRETATION AND DISCUSSION

All species at Lod nice, other than A. konincki, show stablenumbers of segments in adulthood (Fig. 4), demonstratingthat tight control of thoracic segmentation was operative inthis environment. The presence of this constraint in the proe-tide trilobites Scharyia wenlockiana, Decoroproetus coderus,and Thebanaspis putzkeri, phylogenetically intermediate be-tween A. konincki and its Cambrian homoemorphs, demon-

e

strates that relaxation of thoracic constraint was a feature ofsecondarily convergence in A. konincki, and suggest that itwas linked to acquisition of the olenimorphic form. Pheneticanalysis shows that although S. wenlockiana, D. coderus, andT. putzkeri are clearly distinct from one another, they form aphenetic cluster (Figs. 1.2, 1.6, 2.3, 5A, and 6) that is charac-terized by a relatively reduced thorax and enlarged pygidium.This morphotype characterizes almost all the proetides phy-logenetically intermediate between A. konincki and its Cam-

Fig. 7. Provisional ingroup relationships of selected taxa in the order Proetida, based on phylogenetic work in progress, illustrating thegreat phylogenetic distance between Cambrian olenimorphs and Aulacopleura konincki. See Fig. 5B for additional characters. Cambriantrilobites are so labeled, all others are post-Cambrian. 1. Order Proetida: protaspis with preglabellar field. 2. Suborder Aulacopleurina:aulacopleuroid protaspis with paired cephalic tubercles. 3. Pygidium with prominent fulcral tubercles or spines. 4. Superfamily Dimero-pygoidea: yoked librigenae (at least in early ontogeny). 5. Superfamily Bathyuroidea: large, hemicylindrical glabella, strongly vaulted,robust exoskeleton, thick cuticle. 6. Superfamily Aulacopleuroidea: isolated L1. 7. Family Aulacopleuridae: micropygous, thoracic axialspine (primitively). 8. Family Rorringtoniidae: Very small eye with densely packed lenses, entirely lacking socle or platform and set di-rectly on libringenal field. 9. Family Scharyiidae: subtriangular glabella, tail nearly as long as wide, approaching isopygous, cedariformsuture. Generic abbreviations are as in the legend to Fig. 3.

Hughes et al. Trilobite segmentation constraint 33

brian homeomorphs, and forms with this morphotype alsoshow stable numbers of thoracic segments in adulthood (Fig.4). The prediction that relaxation of thoracic constraint is tiedto acquisition of the olenimorphic morphotype is supportedby the pattern of intracollectional, intraspecific variation inthe aulacopleurid “Otarion diffractum” (the species-level tax-onomy of this species is currently in debate), which shows be-tween 17 and 20 thoracic segments in adulthood (Fig. 8). Thisspecies occurs in a thin unit of shales about 5 million yearsyounger than the Lod nice assemblage. As at Lod nice, flex-ibility in segmentation in this olenimorph contrasts with thepattern of stability seen in the co-occurring odontopleurideAcanthalomina minuta, reinforcing the idea that flexibility istaxon specific. These observations confirm that the long-termsecular trend toward stable segment numbers in proetidetrilobites cannot be due to irreversible genetic entrenchmentbecause secondarily convergent olenimorphs could relaxconstraints upon segment numbers when occupying appro-priate niches, despite having sister taxa with tightly con-strained numbers of thoracic segments.

Cambrian trilobite species known to show variation inthoracic segment numbers are not restricted to olenimorphicforms, but include a wide array of morphotypes including thedisparate olenelloid, redlichiid, agnostiid, and primitive lib-ristomate morphologies. Most, but not all, of these formscharacterized by large numbers of thoracic segments inadulthood. However, variation in thoracic segment count inadult trilobites is a rare phenomenon, even in Cambrian tri-lobites—hundreds of species are known whose adults showconstant numbers of segments within bedding plane collec-tions. A detailed review of postcephalic segmentation pat-terns in trilobites is currently in progress, but initial results

e e

reveal that of 20 trilobite species showing well-documentedintracollectional variation in thoracic segment numbers (Ta-ble 1), only two species have less than an average of 11.5segments in adulthood. The degree of variation in segmentnumbers correlates with mean intraspecific numbers of tho-racic segments (r 5 0.424; p , 0.05; n 5 20), so the tempo-ral trend in reduced thoracic variability may be a correlate ofthe trend toward reduced segment number, and a by-productof whatever factors promoted segment-rich thoraces in earlytrilobites. Even so, the apparent possession of either two orthree segments in species of Pagetia (Jell 1975) provides anintriguing exception to the general pattern of stability in seg-ment-poor forms, and a more comprehensive analysis is nec-essary.

The reduction of variability with reduced thoracic seg-ment number could reflect selection for more constant num-bers in forms with fewer segments (Hughes and Chapman1995), or simply be a consequence of the developmentalmechanisms by which segments are specified. The positivecorrelation between the number of segments and degree ofvariation in segment numbers seen in trilobites is also appar-ent among some extant arthropods and vertebrates (Arthur1999; Bateson 1894; Minelli and Bortoletto 1988). We haveno clear idea why there should be a general reduction in tri-lobite thoracic segment numbers with time (possibilities in-clude fewer molts and/or shorter time to sexual maturity),but in trilobites this trend correlates with an increase in thenumbers of segments in the pygidium (Stubblefield 1959).Furthermore, in forms with fewer thoracic segments, themorphological distinction between thoracic and pygidialsegments often increases. Understanding the reasons for theincreasing distincitiveness of the pygidium during trilobite

Fig. 8. Relationship of number of thoracicsegments to overall size, as represented bythe length of the glabella, among the aulaco-pleurid olenimorph “Otarion diffractum”(squares) and the odontopleuride Acantha-lomina minuta (triangles). Note marked in-traspecific variation in segment numbers inthe olenimorph and stability in the co-occur-rent odontopleuride. These fossils occur inbuff calcareous shales at the top of the Cro-mus beaumonti biozone of late Ludlovianage in the south part of Kosov Quarry (solidsymbols) and from “Amerika” quarry nearMo ina (open symbols) (K í 1992). Fig-ured “Otarion diffractum” in upper rightlacks free cheeks on cephalon, and posteriorof pygidium is missing, CGU “JK1” fromKosov Quarry. Figured Acanthalomina mi-nuta in lower left also lacks free cheeks,CGU “JK2” from Kosov Quarry. Scale barrepresents 2 mm in both specimens, and ar-row marks the thoracic/pygidial boundary.

r r z

34 EVOLUTION & DEVELOPMENT Vol. 1, No. 1, July–Aug. 1999

evolution, and its homologies with structures in other arthro-pod taxa, provide major challenges for understanding the de-velopmental history of trilobite body plans.

IMPLICATIONS

The role of developmental constraints in shaping evolutionis of broad interest to evolutionary biologists (see Raff1996), and paleontologists have interpreted the transition be-tween intervals of faster and slower evolutionary change toreflect the evolution of such constraints (e.g., Gould 1989).Empirical evidence for the action of developmental con-straints is sparse (Raff 1996), partly due to the difficulty ofdiscriminating between developmental constraints, whichare intrinsic properties of developmental programs, and ex-trinsic constraints which are determined by an organism’secological setting (Wagner 1995). In the case of trilobites,derived taxa have relatively fixed thoracic segment numberswhen compared to basal forms, and this observation has ledto the suggestion that developmental processes “hardened”or became more entrenched during the evolution of thegroup (McNamara 1983; McKinney and McNamara 1991).However, the shift toward stable numbers of segments mayrather reflect the evolution of derived morphotypes withmore efficient enrollment mechanisms. Enrollment was a be-havioral response to deteriorating environmental conditions,and a defense against predators. In derived groups enroll-ment was a precise and complex procedure that requiredtight regulation of body proportions, including constantnumbers of segments. Hence the trend in segment numbersmay reflect selection for more efficient enrollment but is not,in itself, evidence of the evolution of an intrinsic develop-mental constraint. Such developmental constraint could beinvoked only if the derived pattern of segment stability hadbeen maintained after reversion to a basal morphotype, andis not the case in the example presented above. Analyses ofdevelopmental constraints in fossil groups thus require spe-cial care in selecting well-preserved taxa that provide an ap-propriate test of the constraint model.

It has been argued that post-Cambrian arthropods show arestricted set of segmentation patterns compared to the rangepresent among Cambrian arthropods, and this restriction hasbeen attributed to the evolution of developmental entrench-ment in advanced groups (Gould 1989, 1993; Jacobs 1990).If entrenchment were the cause of this pattern, we should ex-pect evidence for the canalization of segmentation patternswithin each of the major arthropod subclades, restrictingtheir abilities to evolve disparate new body plans. Our resultsdisprove the notion of the evolution of inviolable develop-mental constaints on thoracic segmentation within a well-represented early arthropod subclade. Hence we question theidea that a general increase in developmental constraint after

the early Cambrian radiation was responsible for curtailingorigination of new phylum-level body plans [reviewed inValentine (1995) and Velentine et al. (1996)].

We thank the National Geographic Society for field support of N.C. H. and J. K í . The Czech Geological Survey, Czech NationalMuseum, National Museum of Natural History, and Museum ofComparative Zoology, Harvard University, kindly provided accessto collections in their care and field equipment. R. A. Fortey of theNatural History Museum, London, made loan material available,and provided helpful comments throughout this work. J. K í pro-vided essential expertise in the field. D. E. G. Briggs, M. L. Droser,D. H. Erwin, C. R. Marshall, A. I. Miller, L. H. Smith, and M. Web-ster kindly commented on the manuscript. D. K. Jacobs and C. R.Marshall provided astute reviews for the journal, and R. A. Raffsaw the article through review with great efficiency.

REFERENCES

Adrain, J. M., and Chatterton, B. D. E. 1993. A new rorringtoniid trilobitefrom the Ludlow of Arctic Canada. Canadian J. Earth Sci. 30:1634–1643.

Akam, M., Averof, M., Castelli-Gair, J., Dawes, R., Falciani, F., and Ferri-er, D. 1994. The evolving role of Hox genes in arthropods. Develop-ment (suppl.):209–215.

Arthur W. 1997. The Origin of Animal Body Plans. Cambridge UniversityPress, Cambridge.

Arthur, W. 1999. Variable segment numbers in centipedes: population ge-netics meets evolutionary developmental biology. Evolution and De-velopment 1:62–69.

Bateson, W. 1894, Materials for the Study of Variation. Macmillan, London.Bergström, J., and Levi-Setti, R. 1978. Phenotypic variation in the Middle

Cambrian trilobite Paradoxides davidis Salter at Manuels, S. E. New-foundland. Geol. Palaeontol. 12:1–40.

Bright, R. C. 1959. A paleoecologic and biometric study of the MiddleCambrian trilobite Elrathia kingii (Meek). J. Paleontol. 33:83–98.

Carroll, S. B. 1995. Homeotic genes and the evolution of arthropods andchordates. Nature 376:479–485.

Chapman, R. E. 1990. Conventional Procrustes Approaches. In F. J. Rohlfand F. L. Bookstein (eds.). Proceedings of the Michigan MorphometricsWorkshop. University of Michigan Museum of Zoology, Ann Arbor,MI. Pp. 251–267.

Chatterton, B. D. E., and Speyer, S. E. 1997. Ontogeny. In H. B. Whitting-ton (ed.). Treatise on Invertebrate Paleontology, part O, Arthropoda 1.Trilobita, revised. Geological Society of America and University ofKansas, Boulder and Lawrence, KS. Pp. 173–247.

Chatterton, B. D. E., and Ludvigsen, R. 1998. Upper Steptoean (UpperCambrian) trilobites from the McKay Group of southeastern BritishColumbia, Canada. Journal of Paleontology, Paleontological SocietyMemoir 72 (suppl.):1–43.

Cisne, J. L. 1981. Triarthrus eatoni (Trilobita): anatomy of its exoskeletal,skeletomuscular and digestive systems. Palaeontograph. Am. 9:99–141.

Edgecombe, G. D., Chatteron, B. D. E., Vaccari, N. E., and Waisfeld, B. G.1997. Ontogeny of the proetoid trilobite Stenoblepharum, and relation-ships of a new species from the Upper Ordovician of Argentina. J. Pa-leontol. 71:419–433.

Fortey, R. A. 1990. Ontogeny, hypostome attachment and trilobite classi-fication. Palaeontology 33:529–576.

Fortey, R. A. 1997. Classification. In H. B. Whittington (ed.). Treatise onInvertebrate Paleontology, part O, Arthropoda 1. Trilobita, revised.Geological Society of American and University of Kansas, Boulder andLawrence, KS. Pp. 289–302.

Fortey, R. A., and Chatterton, B. D. E. 1988. Classification of the trilobitesuborder Asaphina. Palaeontology 31:165–222.

Fortey, R. A., and Owens, R. M. 1975. Proetida—a new order of trilobites.Foss. Strata 4:227–239.

r z

r z

Hughes et al. Trilobite segmentation constraint 35

Fortey, R. A., and Owens, R. M. 1990. Trilobites. In K. J. McNamara (ed.).Evolutionary Trends. Belhaven Press, London. Pp. 121–142.

Gould, S. J. 1989. Wonderful Life. Hutchinson Radius, London.Gould, S. J. 1993. How to analyze Burgess Shale disparity—a reply to Rid-

ley. Paleobiology 19:522–523.Gunther, L. F., and Gunther, V. G. 1981. Some Middle Cambrian fossils of

Utah. Brigham Young University Geology Studies 28:1–87.Henningsmoen, G. 1952. Early Middle Cambrian fauna from Rogaland,

SW Norway. Norsk Geologisk Tidsskrift 30:13–32.Henningsmoen, G. 1957. The Trilobite family Olenidae. Skrifter Utgitt av

Det Norske Videnskaps—Akademi i Oslo 1. Matematisk—Naturviden-skabelig Klasse 1:1–303.

Hughes, N. C., and Chapman, R. E. 1995. Growth and variation in the Sil-urian proetide trilobite Aulacopleura konincki and its implications fortrilobite palaeobiology. Lethaia 28:333–353.

Jacobs, D. K. 1990. Selector genes and the Cambrian radiation of Bilateria.Proc. Natl. Acad. Sci. USA 87:4406–4410.

Jell, P. A. 1975. Australian Middle Cambrian eodiscoids with a review ofthe superfamily. Palaeontographica, Abt. A 150:1–97.

K í , J. 1992. Silurian Field Excursions: Prague Basin (Barrandian), Bohe-mia. National Museum of Wales, Geological Series No. 13:1–111.

McKinney, M. L., and McNamara, K. K. 1991. Heterochrony, the Evolu-tion of Ontogeny. Plenum Press, New York.

McNamara, K. J. 1983. Progenesis in trilobites. Special Papers in Palaeon-tology 30:59–68.

McNamara, K. J. 1986. The role of heterochrony in the evolution of Cam-brian trilobites. Biol. Rev. 61:121–156.

Minelli, A., and Bortoletto, S. 1988. Myriapod metamerism and arthropodsegmentation. Biol. J. Linn. Soc. 33:323–343.

P ibyl, A. 1947. Aulacopleura and the Otarionidae. J. Paleontol. 21:537–545.

Raff, R. A. 1996. The Shape of Life. Genes, Development, and the Evolu-tion of Animal Form. University of Chicago Press, Chicago, IL.

r z

r

Robison, R. A. 1971. Additional Middle Cambrian trilobites from theWheeler Shale of Utah. J. Paleontol. 45:796–804.

Sadler, P. M., and Strauss, D. J. 1990. Estimation of completeness of strati-graphical sections using empirical data and theoretical models. J. Geol.Soc. Lond. 147:471–485.

Sassaman, C., Simovich, M. A., and Fugate, M. 1997. Reproductive isola-tion and genetic differentiation in North American species of Triops(Crustacea: Branchiopoda: Notostraca). Hydrobiologia 359:125–147.

Schram, F. R. 1986. Crustacea. Oxford University Press, New York.Schwimmer, D. R. 1989. Taxonomy and biostratigraphic significance of

some Middle Cambrian trilobites from the Conasauga Formation inwestern Georgia. J. Paleontol. 63:484–494.

Sommer, R. J., and Tautz, D. 1993. Involvement of an orthologue of theDrosophila pair-rule gene hairy in segment formation of the short germ-band embryo of Tribolium (Coleoptera). Nature 361:448–450.

Stubblefield, C. J. 1959. Evolution of trilobites. J. Geol. Soc. Lond.115:145–162.

Thomas, A. T., and Lane, P. D. 1984. Autecology of Silurian trilobites. Spe-cial Papers Palaeontol. 32:55–69.

Valentine, J. W. 1995. Why no new phyla after the Cambrian? Genome andecospace hypotheses revisited. Palaios 10:190–194.

Valentine, J. W., Erwin, D. H., and Jablonski, D. 1996. Developmentalevolution of metazoan bodyplans: the fossil evidence. Dev. Biol.173:373–381.

Wagner, P. J. 1995. Testing evolutionary constraint hypotheses: exampleswith early Paleozoic gastropods. Paleobiology 21:248–272.

Wang, Q.-Z. 1989. Early Silurian trilobites from Wulong, southeastern Si-chuan (China). J. Hebei College of Geology 12:422–440.

Westergård, A. H. 1936. Paradoxides oelandicus beds of Öland. SverigesGeologiska Undersökning C394:1–66.

Whittington, H. B. 1989. Olenelloid trilobites: type species, functional mor-phology and higher classification. Philos. Trans. R. Soc. Lond. B324:111–147.