|*B& ’wWSumF*' ■ ^ w 4 ‘• - zk,y '.:. v > ? .. st** t' -7* f^WLy-;^ MU 1 111 gj ■ nTnjfiCTi H ■ 1 MtJrPffjdr* jJEdWSF'. .JKT j&m i yf - L 1 f j HI. Jjl jV l Is Vl IV vSS B*4 > l^‘. *v 2s? «;**^K Ifl w(l ffio>JB W ■# M M^Kj y^ 'j, ^ Vlj *-^£5S2Ik'£V* SERIES PUBLICATIONS OF THE SMITHSONIAN INSTITUTION Emphasis upon publication as a means of "diffusing knowledge" was expressed by the first Secretary of the Smithsonian. In his formal plan for the institution, Joseph Henry outlined a program that included the following statement: "It is proposed to publish a series of reports, giving an account of the new discoveries in science, and of the changes made from year to year in all branches of knowledge.” This theme of basic research has been adhered to through the years by thousands of titles issued in series publications under the Smithsonian imprint, commencing with Smithsonian Contributions to Knowledge in 1848 and continuing with the following active series: Smithsonian Contributions to Anthropology Smithsonian Contributions to Botany Smithsonian Contributions to the Earth Sciences Smithsonian Contributions to the Marine Sciences Smithsonian Contributions to Paleobiology Smithsonian Contributions to Zoology Smithsonian Folklife Studies Smithsonian Studies in Air and Space Smithsonian Studies in History and Technology In these series, the Institution publishes small papers and full-scale monographs that report the research and collections of its various museums and bureaux or of professional colleagues in the world of science and scholarship. The publications are distributed by mailing lists to libraries, universities, and similar institutions throughout the world. Papers or monographs submitted for series publication are received by the Smithsonian Institution Press, subject to its own review for format and style, only through departments of the various Smithsonian museums or bureaux, where the manuscripts are given substantive review. Press requirements for manuscript and art preparation are outlined on the inside back cover. Robert McC. Adams Secretary Smithsonian Institution SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY • NUMBER 79 Ontogeny, Intraspecific Variation, and Systematics of the Late Cambrian Trilobite Dikelocephalus Nigel C. Hughes SMITHSONIAN INSTITUTION PRESS Washington, D.C. 1994 ABSTRACT Hughes, Nigel C. Ontogeny, Intraspecific Variation, and Systematics of the Late Cambrian Trilobite Dikelocephalus. Smithsonian Contributions to Paleobiology, number 79,89 pages, 47 figures, 11 plates, 27 tables, 1994.—Biometric analyses of well-localized specimens of the trilobite Dikelocephalus from the St. Lawrence Formation (Upper Cambrian), northern Mississippi Valley, suggest that all specimens belong to a single, highly variable morphospecies, D. minnesotensis. A complex pattern of ontogenetically-related and ontogeny-independent variation produced a mosaic of morphotypes, which show greater diversity than previously recorded within trilobite species. There is considerable variation within collections made from single beds. Variations of characters among collections are mosaic, and are clinal in some cases. Patterns of variation within Dikelocephalus cannot be related to lithofacies occurrence. There are no obvious temporal variations in D. minnesotensis within the St. Lawrence Formation, but some Dikelocephalus from the underlying Ttinnel City Group may belong to a different taxon. The validity of this early taxon is questionable due to a lack of available material. The mosaic pattern of variation in Dikelocephalus mimics that documented at higher taxonomic levels in primitive libristomate trilobites, and helps explain difficulties in providing a workable taxonomy of primitive trilobites. Results caution proposition of evolutionary scenarios that do not take account of intraspecific variation. The recovery of dorsal shields of Dikelocephalus permits the first detailed reconstruction of the entire exoskeleton. The systematics of the genus is revised and twenty-five species are suppressed as junior synonyms of D. minnesotensis. Official publication date is handstamped in a limited number of initial copies and is recorded in the Institution’s annual report, Smithsonian Year. Series COVER DESIGN: The trilobite Phacops rana Green. Library of Congress Cataloging-in-Publication Data Hughes, Nigel C. Ontogeny, intraspecific variation, and systematics of the Late Cambrian trilobite Dikelocephalus / Nigel C. Hughes p. cm. - (Smithsonian contributions to paleobiology ; no. 79) Includes bibliographic references. 1. Dikelocephalus. 2. Paleontology-Cambrian. I. Title. II. Series. QE701.S56 no. 79 [QE823.P79) 560 s-dc20 [562'.393] 94-28605 ® The paper used in this publication meets the minimum requirements of the American National Standard for Permanence of Paper for Printed Library Materials Z39.48—1984. Contents Page Introduction. 1 Geological Setting. 1 Previous Work. 2 Approach and Methodology. 4 Approach. 4 Species Concept. 6 Methods. 6 Materials . 6 Statistics. 7 Analysis of Nominal and Ordinal Characters. 8 Cranidium. 8 Anterior Border. 8 Caeca. 9 Eye Ridges.10 Eye Position.10 Ornamentation.10 Glabella.10 Anterior Margin.10 Lateral Glabellar Furrows.10 Median Occipital Tlibercle.11 Ornamentation.12 Free Cheeks.13 Median Suture.13 Lateral Border.15 Length of Genal Spine.15 Ornamentation.15 Hypostome (Labrum).15 Thorax.16 Number of Thoracic Segments.16 Anterior Margin of Thoracic Segments.17 Ornamentation.17 Pygidium.17 Axial Furrows.17 Termination of the Axis.17 Post-axial Ridge.17 Post-axial Emargination.17 Number of Pleural and Interpleural Furrows.17 Division of the Pleurae.18 Length of the Pleurae.19 Marginal Rim.21 Ornamentation.21 Discussion.22 Bivariate Analysis.23 Introduction.23 Approach.23 Methods.23 in SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Results.24 Cranidium .24 Glabellar Width.24 Frontal Area Length.26 Frontal Area Width.28 Palpebral Lobe Length.31 Pygidium. 33 Pygidial Width.33 Pygidial Axial Length.33 Posterolateral Spine.35 Discussion.39 Growth Controls.39 Population-related Variation.39 Intraspecific Heterochrony.40 Tunnel City Group Dikelocephalus .40 Character Independence.41 Multivariate Analysis.41 Introduction.41 Multivariate Methods.41 Principal Component Analysis.41 Nonmetric Multidimensional Scaling.42 Results.42 Cranidia.42 Pygidia.45 Discussion.51 Systematic Paleontology.53 Morphological Variation in Dikelocephalus .58 Introduction.58 Variation Controls.58 Growth-related Variation.58 Population-related Variation.58 Conclusions.60 Acknowledgments.60 Literature Cited.61 Plates.67 Ontogeny, Intraspecific Variation, and Systematics of the Late Cambrian Trilobite Dikelocephalus Nigel C. Hughes Introduction Dikelocephalus is a large asaphide trilobite that is abundant in Upper Sunwaptan ( Saukia Zone) deposits of the northern Mississippi Valley (central to western Wisconsin, southeastern Minnesota, and northeastern Iowa) (Hughes, 1993). Dikelo¬ cephalus grew to a length of 40 cm or more and it has often been used as a textbook example of a Cambrian trilobite. In spite of this, the genus remains poorly described and no detailed reconstruction has ever been presented. The taxonomy of Dikelocephalus also remains poorly resolved and none of the wide variety of taxonomic schemes proposed (Owen, 1852; Walcott, 1914; Ulrich and Resser, 1930; Twenhofel, 1945; Rassch, 1951; Labandeira, 1983) has proved satisfactory. This paper is a comprehensive description of the morphology, intraspecific variation, and systematics of D. minnesotensis from the St. Lawrence Formation of Wisconsin, Minnesota, and Iowa. It is one of a series of papers that encompass the geological setting, taphonomy, population paleobiology, tax¬ onomy, and functional morphology of the species (Hughes, 1990, 1991, 1993; Labandeira and Hughes, 1994). This paper provides documenation for the remarkable pattern of variation within Dikelocephalus minnesotensis Owen that has been briefly summarized elsewhere (Hughes, 1991). It also discusses the broader implications of these results for trilobite sys¬ tematics and for questions about evolutionary mechanisms operating in the Cambrian. Geological Setting During the Late Cambrian three broad lithofacies belts surrounded the exposed Laurentian shield (Palmer, 1960, 1965). Each of these lithofacies belts supported a distinctive Nigel C. Hughes, Cincinnati Museum of Natural History, 1720 Gilbert Ave., Cincinnati, Ohio 45202 (Research Associate, Department of Paleobiology, National Museum of Natural History, Smithsonian Institution). range of trilobite biofacies (Taylor, 1977; Ludvigsen and Westrop, 1983a). Species of the family Dikelocephalidae inhabited shelf seas across much of North America during the latest Cambrian Sunwaptan Stage (Ludvigsen and Westrop, 1985), but the genus Dikelocephalus is mainly restricted to the inner-shelf mixed carbonate/siliciclastic belt. Its remains are found in large numbers only in the northern Mississippi Valley (in the sense of Dott et al., 1986) (central to western Wisconsin, southeastern Minnesota, and northeastern Iowa). This region is bounded to the south by the carbonate-dominated lithofacies, with distinct suites of trilobites, and to the north and west by the exposed Laurentian shield. This paleogeographic setting makes Dikelocephalus a suitable candidate for analysis of morphol¬ ogic variation occuring within a single basin, as patterns of variation are unlikely to be affected by immigrations from other regions. During the last three million years of the Cambrian Period Dikelocephalus was abundant and widely distributed through¬ out the northern Mississippi Valley (Figure 1). The oldest known Dikelocephalus were collected from the upper part of the Thnnel City Group (Figure 2). It occurs in both the very fine-grained, glauconite-rich feldspathic sandstones of the Reno Member (the uppermost unit of the Lone Rock Formation) and in its correlative nearshore facies, a fine¬ grained, non-glauconitic sandstone called the Mazomanie Formation (for discussion of lithostratigraphy see Ostrom, 1978). Dikelocephalus specimens are found most common within the St. Lawrence Formation, which directly overlies the Tlinnel City Group (Figure 2). The principal facies in this formation is heterolithic, consisting of fine-grained clay-rich dolostones that are finely interbedded with feldspathic sand¬ stones of very fine grain size. Specimens are occasionally found in laminated sandstone facies, which occur nearer to the paleoshoreline. The youngest known Dikelocephalus are fragments collected from medium-grained sandstones of the Van Oser Member of the Jordan Formation, which overlies the St. Lawrence Formation. Dikelocephalus occurs in the Saukia 1 2 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 1.—Location of major collections of Dikelocephalus minnesotensis Owen from the St Lawrence Formation. Key collections are those analyzed using multivariate statistical techniques. Approximate outcrop area of northern Mississippi Valley Cambrian deposits is shown in the top right comer. Full locality details are given in Hughes (1993). Zone (in the sense of Raasch, 1951) of the Upper Sunwaptan Stage (Ludvigsen and Westrop, 1985), also known as the Trempealeauean Stage. Details of the stratigraphic and taphon- omic distribution of Dikelocephalus in the northern Mississippi Valley, along with discussions of the mode of preservation and its effects on taxonomy, functional morphology, and paleoecol- ogy are provided by Hughes (1990, 1993) and Labandeira and Hughes (1994). These analyses have shown that the spatial and temporal distribution of Dikelocephalus from the northern Mississippi Valley is well understood and that the present distribution of specimens reflects the primary biological distribution of Dikelocephalus. Results suggest that Dikelo¬ cephalus from this region are suitable candidates for a detailed locality-based study. Previous Work The genus Dikelocephalus when first described by Owen (1852) included a broad range of libristomate trilobites (in the sense of Fortey, 1990). None of Owen’s species, other than D. minnesotensis, are now assigned to the same family as Dikelocephalus. A series of works, based mainly on material collected subsequent to Owen’s work, (Hall, 1863; Winchell, 1874) established the basis of the family Dikelocephalidae (Miller, 1889). Following Walcott’s (1914) revision of the genus, Ulrich and Resser (1930) produced an extensive monograph in which twenty-six species of Dikelocephalus were described from the northern Mississippi Valley. This monograph has been criticized mainly on the grounds that the taxonomy of the group was subdivided too finely (Twenhofel, 1945; Raasch, 1951; Taylor and Halley 1974; Westrop, 1986). Despite the criticism there have been few attempts to reassess the status of Dikelocephalus species and additional species have been described (Bell, Feniak, and Kurtz, 1952). Raasch (1951) attempted a revision of dikelocephalid and saukiid trilobites, but this work consisted of a list of synonymies with little discussion and without illustrations. Raasch’s (1951) revision also relied heavily on stratigraphic rather than morphological criteria. Dikelocephalus has since been recorded in other areas of north America (Grant, 1965; Winston and NUMBER 79 3 Figure 2.—Stratigraphic sections of nine localities from the northern Mississippi Valley. The sections lie along an approximate northwest-southeast transect Occurrence of Dikelocephalus minnesotensis Owen is indicated by large stars. Closed stars indicate collections that have been analyzed using multivariate statistics. Open stars record the occurrence of D. minnesotensis at this horizon. Full locality details are given in Hughes (1993). 4 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Nichols, 1967; Stitt, 1971; Westrop, 1986) but specimens have been usually identified only to the generic level. Ulrich and Resser’s (1930) approach to the systematics of dikelocephalid trilobites focused on character variability; almost any feature that showed morphological variation was automatically considered to be of taxonomic importance. The result was a large number of species descriptions based on minor and inconsistent differences between specimens. Later workers (e.g., Twenhofel, 1945) found this taxonomy impossi¬ ble to apply to new material. Using the same suite of specimens that Ulrich and Resser (1930) described, Labandeira (1983) and Labandeira and Hughes (1994) quantitatively assessed the validity of these species designations. The results suggest that all specimens form part of a single morphospecies. Ulrich and Resser (1930) also used some meristic characters in their species diagnoses. The distribution of some of these characters could potentially serve to define species within Dikelocephalus. The significance of these characters has not been previously re-assessed. Finally, Ulrich and Resser’s (1930) type suite was made up of material collected from a large number of localities. Many of these localities were represented by one or two specimens, and some of these specimens possess characters which could be species diagnostic. It is not possible to assess the taxonomic significance of these characters without understanding the range of variation within collections from single bedding planes. Hence it is impossible to resolve the taxonomic status of Dikelocephalus on the basis of Ulrich and Resser’s type-suite alone (Labandeira and Hughes, 1994). The present study concentrates on new and well-localized collections to determine variation within Dikelocephalus and aims to resolve those characters that can be used to recognize discrete taxa. A list of characters considered important for species recognition by previous authors (Table 1) documents a wide range of morphological variation within Dikelocephalus. Most of Ulrich and Resser’s (1930) species descriptions were written as comparisons, rather than as diagnoses. These comparsions were not comprehensive, with the result that Ulrich and Resser’s species descriptions are difficult to evaluate. Hence many of the characters listed in Table 1 were selected by observation of the type specimens of Ulrich and Resser’s (1930) species, rather than from their original descriptions. Approach and Methodology Approach The contrasting taxonomic schemes of Ulrich and Resser (1930), Raasch (1951), and Labandeira (1983) are indicative of the remarkable nature of variation in Dikelocephalus. Early taxonomic splitting highlighted the marked degree of variation in some characters (Ulrich and Resser 1930; Raasch, 1951), whereas bivariate analyses illustrated that many other charac¬ ters are relatively invariant (Labandeira, 1983). In order to fully document the morphology of Dikelocephalus a more compre- Table 1.—Morphological characters of previously described species and varieties herein assigned to Dikelocephalus. The characters were selected from published descriptions and from inspection of figured type material. See text for discussion. Hall (1863) variety limbatus border present caeca present Ulrich and Resser (1930) minnesotensis border slightly inflated frontal area wide medial portion of glabellar anterior straight angle of lateral glabellar furrows to sagittal axis acute subequally divided pygidial pleurae barretti glabella and pygidial axis pustulated pygidium rounded beani pygidial pleurae unequally divided brevis palpebral lobes large frontal area very short posterolateral spines long pygidial pleurae equally divided declivis median suture absent genal spines short free cheeks flat edwardsi caeca present border present median suture absent gracilis frontal area short maculae present median suture absent posterolateral spines long pygidial pleurae equally divided granosus cephalon granulated palpebral lobes of high relief glabellar furrows firmly impressed halli frontal area wide pygidial pleurae subequally divided hotchkissi medial portion of glabellar anterior straight glabellar sides convergent palpebral lobe small angle of lateral glabellar furrows to sagittal axis obtuse free cheeks fiat pygidial pleurae subequally divided inaequalis pygidial pleurae unequally divided intermedicus frontal area long, fiat frontal area anterior margin rounded fifth pygidial axial ring present pygidial pleurae subequally divided NUMBER 79 5 juvinalis small size frontal area short, wide marginatus cranidial border present median suture absent pygidial furrows converge at base of posterolateral spine norwalkensis cranidial border present frontal area anterior margin rounded glabellar furrows firmly impressed post-axial ridge present barretti cranidial border present faint caeca frontal area short median suture absent equally divided pleurae posterolateral spines long inaequalis pygidial pleurae unequally divided marginatus cranidial border present pygidial furrows converge at base of posterolateral spine orbiculatus cranidial margin rounded palpebral lobes large pygidium rounded oweni no shared characters, separating this taxon from others, are apparent among Ulrich and Resser's (1930) type specimens that Raasch assigned to D. oweni ovatus glabella wide pygidium rounded oweni anterior margin of frontal area angular frontal area short palpebral lobes large pygidial pleurae unequally divided postrectus pygidial posterior margin straight raaschi frontal area narrow, flat palpebral lobes large median suture present pygidial pleurae unequally divided retrorsus frontal area short occular platform narrow posterolateral spine long subplanus frontal area anterior margin rounded median suture absent pygidium elongate thwaitesi cranidial border present free cheeks wide weidmani border present maculae absent pygidial pleurae unequally divided wiltonensis border present glabella anterior margin rounded frontal area anterior rounded wisconsinensis frontal area anterior rounded palpebral lobes large genal spine long pygidium rounded Raasch (1951) minnesotensis frontal area long, wide palpebral lobes small pygidial pleurae subequally divided postrectus pygidial posterior margin straight thwaitesi cranidial border present frontal area long unequally dividied pleurae hensive approach is necessary. One possiblity would be a multivariate investigation of the entire suite of Dikelocephalus specimens. This approach was rejected because of (1) the fragmented condition of the majority of material collected previously, (2) the difficulty of interpreting the results of multivariate analyses based on “mixed-mode” data sets (those including both metric and non-metric characters), and (3) many collections with sample sizes too small to assess the degree of intraspecific variation. To overcome these problems the analysis of variation in Dikelocephalus reported here proceeded in three stages. First, variations in nominal (presence/absence) and ordinal (ranked scale) characters were analyzed using specimens from through¬ out the northern Mississippi Valley. These characters were analyzed individually to permit recognition of their distinct patterns of variation within and among collections. Analyses did not reveal any consistent variations within the whole sample. Several seemingly discrete characters exhibit continu¬ ous variation. Some characters show size-dependent variation, and there is great morphological variability within collections from individual beds. Having failed to detect consistent discontinuous variation within Dikelocephalus the resolution of the morphological analysis was increased. Quantitative analyses of collections from individual beds were used to examine details of holaspid growth and assess population-related variation. Bivariate analyses show significant allometry in holaspid growth. Patterns of intra- and intercollection variability are continuous and highly complex. The independence of many characters, combined with subtle population-related differences in growth rates, evidently results in a mosaic of morphologies, indicating high levels of plasticity within Dikelocephalus. The third stage of the work was to examine whether there are 6 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY subtle patterns of covariance within the data set that were undetected in bivariate analyses. Multivariate analyses did not indicate morphological discontinuity within Dikelocephalus. Rather, it confirms the results of the other approaches, suggesting that many characters within Dikelocephalus vary independently. Species Concept The species concept employed in this study is “the smallest aggregation of populations ... diagnosable by a unique combination of character states in comparable individuals” (Nixon and Wheeler, 1990:218). Such phylogenetic species cannot be further subdivided even if they show considerable continuous variation within the lineages (Wiley, 1978). Methods 1. Analysis of nominal and ordinal characters. Some morphological variation within Dikelocephalus reflects varia¬ tions in discrete characters, either presence/absence characters or characters that occur in a small number of discrete states. These are referred to as nominal characters. Other characters vary across a continuous spectrum but their exact position within this spectrum cannot be assessed quantitatively. Instead, only their relative or ranked position can be determined. These are called ordinal characters. Nominal and ordinal characters were analyzed in more than 2750 specimens from 93 localities in the northern Mississippi Valley (detailed in Hughes, 1993). In addition to features previously unrecognized in Dikeloceph¬ alus, analysis of nominal and ordinal characters included all the non-metric characters used by Ulrich and Resser (1930), who used these sorts of features extensively in their taxonomic study. Such characters formed the basis for some of their specific designations but were largely ignored by Labandeira (1983) and Labandeira and Hughes (1994) in their analyses of the type-suite. 2. Bivariate analysis of collections from individual beds. Bivariate analyses are particularly useful for investigating the growth dynamics of sclerite dimensions. Labandeira’s (1983) bivariate study of Ulrich and Resser’s material provided a valuable overview of bivariate relationships but did not examine growth controls or locality-related variation. Because the present study focuses on the growth- and locality-related controls of variation within Dikelocephalus, it is essential to perform detailed locality-based bivariate analyses. 3. Multivariate analysis of collections from individual beds. Multivariate analyses have been used to examine the relation¬ ships among sclerite dimensions, and to assess the overall similiarity among specimens. Multivariate anaylses are com¬ prehensive methods of assessing morphological variation because a large number of characters may be analyized synoptically, although results may be difficult to interpret (Palmer, 1985). In this study multivariate analyses were performed on samples from single beds which contained large numbers of specimens. Multivariate analyses were limited to samples of sufficient size to permit assessment of the range of intra-sample variation. Metric and angular data were included in the multivariate analyses, nominal and ordinal characters were excluded. Materials Morphological variation in Dikelocephalus was assessed from about 1000 specimens collected during this study, combined with specimens held in 10 North American museums and universities (see Hughes, 1993), specimens belonging to Mr. Gerald Gunderson, and the Sauk County Historical Museum. In total, more than 2750 specimens were examined. Collections with excellent locality data were used to analyze variation within and between collection of Dikelocephalus. Studies were concentrated on collections from single beds from six localities, all within the range of the Saukia fauna (see Hughes, 1990). The collections used were: North Freedom Bed 2 (NF2; quarry on west side of Mirror Lake Road, 0.5 km north of intersection with Hwy 136, 4 km north of North Freedom, Sauk County, Wisconsin), North Freedom Bed 8 (NF8; locality as NF2), Arcadia Bed 18 (A Aa; quarry on east side of Hwy 93, 4 km south of Arcadia, Trempealeau County, Wisconsin), Button Bluff Bed 6 of G.O. Raasch (1939) (LRc; quarry above Button Bluff Cemetary, 4 km east of Gotham, Richland County, Wisconsin), J.C. Ferguson’s collection from Stillwater (SWb; roadcut on Hwy 95, 0.5 km north of Stillwater, Washington County, Minnesota), and LaGrange Mountain (RWa; east side of Hwy 61, town of Redwing, Goodhue County, Minnesota) (Figure 1). Full locality details are given in Hughes (1993). With the exception of Stillwater, each of these localities has been the subject of detailed sedimentological analyses by Hesselbo (1987) or Hughes (1990). Each collection contained a large number of well-preserved specimens, which were prepared in the laboratory using standard preparatory techniques. The localities were selected to give a broad geographic coverage of the region (as an example, Stillwater and North Freedom are approximately 300 km apart) (Hughes, 1993, fig. 2), and to illustrate possible temporal variation within the Saukia fauna times. The selection aimed to represent the onshore/offshore proximality trend and to parallel paleo- shore line. Collections from different horizons at the same site were treated as separate statistical populations (i.e., North Freedom Beds 2 and 8; Stillwater a and b). In order to increase temporal resolution, a pooled sample of specimens from the Tbnnel City Group was also include in the morphometric analyses (see Hughes, 1993, fig. 1). Small sample sizes prevented locality-based analyses of Humel City Group Dikelocephalus, but this problem may be limited, because the specimens come from a much smaller geographic area than do samples from the St. Lawrence Formation (Hughes, 1993, fig. 2). Specimens from the TUnnel City Group are older than those NUMBER 79 7 from the St. Lawrence Formation but stratigraphic relationships among the samples within the St. Lawrence Formation are not certain, except in those cases where collections were made from different beds at the same locality. Limited evidence suggests that the northern collections (Stillwater a and b) are older than that of Arcadia (AAa), which is in turn older than collections from North Freedom (NF Beds 2 and 8) (Hughes, 1990). In an exploratory survey of morphological variation sclerites were placed in a Rost planvariograph (see Hughes and Rushton, 1990) and projected to a standard length. The outline of the projected sclerite was drawn onto tracing paper. Several traced images could then be superimposed, allowing swift visual assessment of differences. Some characters appeared static; others showed remarkable variability. Appropriate metric dimensions were then selected for morphometric analysis. Linear dimensions (Table 2; Figure 3) were measured using calipers, angular dimensions with a protractor. Because most specimens were too large to be measured using a microscope, measurements were taken normal to the surface being measured, rather than normal to a standardized plane (Shaw, 1957; Tfemple, 1975a). However, because many of the dimensions were taken from essentially planar surfaces, the disparity between surface-normal measurements and standard orientation measurements is insignificant. Fossil preservation placed limitations on the range of characters available for study. Many sclerites are incomplete, preventing measurement of a full suite of characters. Laban- deira and Hughes (1994) have shown that compaction has limited effect on statistical relationships of cranidia and pygidia of Dikelocephalus, provided that markedly distorted specimens are first removed from the data set. This may be because compaction does not result in marked lateral expansion (Briggs and Williams, 1981; Hughes, 1993), and hence has little effect on the measurements used in these analyses. This interpretation is supported by the observation (see below) that the strongest correlations in Dikelocephalus are between those metric dimensions most likely to be affected by compaction (i.e., orthogonal dimensions on arched surfaces). Furthermore, although many specimens included in these analyses have undergone compaction, the degree of compaction appears to be approximately similar in all specimens. Free cheeks, thoracic segments, and ventral sclerites were excluded from the morphometric study. The curved shape of free cheeks limits the number of homologous points that can used in metric analysis. Free cheeks were also particularly prone to shape distortion during flattening due to their convexity (Hughes, 1993, pi. 5: figs. 4, 5), and such distortion has been shown to be the main cause for the variation in their shape (Labandeira, 1983). Isolated thoracic segments were excluded from the analysis because complete specimens show that segment morphology varies gradually along the thorax (see below), making it impossible to ascertain the original position of any disarticulated segment. TABLE 2.—Symbols used for sclerite dimensions (after Shaw, 1956, 1957; Tfemple, 1975a). Character Abbreviation Frontal area length (sag) fl Frontal area width (tr) j2 Occipital-glabellar length (sag) bl Occipital lobe width (tr) k Palpebral lobe length (exsag) c Intra-articulating pygidial length (sag) zl Intra-articulating pygidial spine length (exsag) z5 Pygidium axial length (sag) yi Maximum pygidial width (tr) w Figure 3. —Linear dimensions measured on cranidia and pygidia of Dikelocephalus. Abbreviations used for sclerite dimensions follow Shaw (1956, 1957) and Tfemple (1975a). For explanation see Table 2. Statistics The Pearson product-moment correlation coefficient and the reduced major axis were calculated in the analyses of nominal and ordinal characters. The product-moment correlation coeffi¬ cient ( r ) measures the degree of association between two variables by assessing the extent to which two variables covary through growth. Comparison of the correlation coefficients of one variable with others allows assessment of the relative variability of that character. 8 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Reduced major axis (RMA) is a form of regression analysis that describes the relationship between two characters during growth. It specifies the initial growth index (b) which is the relationship of the two variables at the start of growth, and the growth ratio (a) which is the ratio between the growth rates of the two variables (Imbrie, 1956). The reduced major axis line is calculated as the line that minimizes the sum of the products of the two variables (Rayner, 1985:425). This technique was used because it gives equal weight to each variable when calculating the best fit line, rather than assuming that one variable is dependant on the other as in conventional regression analysis (Imbrie, 1956). If the ratio between two dimensions of an organism does not change during ontogeny then growth is isometric. If the ratio changes, then growth is allometric (Gould, 1966; Raup and Stanley, 1985). Reduced major axis analysis can be used to assess whether or not the growth relationships of a pair of characters are similar in different populations or taxa. It has been used in many studies of trilobites (e.g., Selwood, 1966; Pabian and Fager- strom, 1968, 1972; Eldredge, 1972; Engel and Morris, 1989) and describes the growth trajectory (Eldredge, 1972) within a population, allowing a series of trajectories from different populations to be compared. In this study the reduced major axes of different collections were compared at the 95% significance level using the method of Imbrie (1956) and tests for isometry were computed using equations and tables given by Hayami and Matsukuma (1970). Statistically significant differences do not in themselves necessarily imply taxonomic differences (e.g., Sheldon, 1987). There are examples of statistically significant differences between populations in sclerite dimensions of some living arthropod species, (e.g., Limulus polyphemus Linnaeus: Riska, 1981 and references therein). If species are defined as basal differentiated taxa each must be characterized by unique combination of character states. Statistically significant differences between populations may be considered unique only if they define mutually exclusive groups the ranges of which do not overlap. Intracollection variation can be assessed by the spread of variation about the generalized growth trend (i.e., reduced major axis). The coefficient of relative dispersion is a measure of this variation and expresses, as a percentage, the amount of shape variation as a proportion of the average shape within the sample (Imbrie, 1956). Analysis of Nominal and Ordinal Characters Cranidium Anterior Border.—A swollen anterior border is common in many specimens of Dikelocephalus from the northern Mississippi Valley. One of Owen’s syntype cranidia from Stillwater, Minnesota (SWa) (Owen, 1852, table 1, fig. 1), shows the development of an anterior border (Plate 11: Figure 17; Labandeira and Hughes, 1994, pi. 1: fig. 1) and many specimens from laminated sandstones show the development of an upturned or inflated anterior border (Plate 11: Figures 6, 7; Labandeira and Hughes, 1994, pi. 1: figs. 4-6). However, not all specimens preserved with the original relief possess a raised border (Hughes, 1993, pi. 2: figs. 3, 4) and the variation of the cranidial anterior margin is striking. The raised border can often be recognized within collections from heterolithic facies (Hughes, 1990). These fine-grained lithologies also commonly preserve terrace lines that run transversely along the anterior border (Plate 2: Figure 3, Plate 3: Figures 2-4, Plate 6: Figure 2; Hughes, 1993, pi. 4: figs. 6-8). The variable nature of the anterior border was studied using collections from four localities (NF2, NF8, AAa, and SWb) from heterolithic facies and one collection from the laminated sandstones of the Tlinnel City Group (FG). Within each of these collections there was a variety of forms, some possessing an anterior border, others without. Variation in border morphology was not linked to other features indicative of flattening, such as glabellar relief. In each of the collections the specimens with relatively long (sag.) frontal areas were those that also possessed an anterior border (Plate 2: Figure 3, Plate 3: Figures 2-4, Plate 4: Figure 4, Plate 6: Figures 6, 8-11, Plate 9: Figures 1-5, 7-11). This suggests that the presence or absence of the border is related to the position of the facial suture. In forms with long frontal areas the anterior of the cranidium probably extended to the anterior margin of the cephalon (as in the complete cephalon illustrated in Plate 1: Figure 1). This is supported by the observation that in many fused free cheeks the facial suture follows the anterior margin of the cephalon (Plate 2: Figure 4, Plate 3: Figures 7, 8; Hughes, 1993, pi. 4: fig. 2). However, in some individuals the anterior section of the facial suture is not marginal, but crosses the dorsal surface at some distance from the cephalic margin. In such cases the anterior of the cranidium is described as “retracted” from the anterior margin of the cephalon (Plate 5: Figure 2; Labandeira and Hughes, 1994, pi. 1: fig. 9). The extent of retraction is variable, but there is no relationship between cephalic size and either extent of retraction or the presence or absence of a raised anterior border (cf., Ludvigsen and Tliffnell, 1983, for Triarthrus ). If marginal and retracted conditions characterize distinct morphs, this should be reflected in the results of statistical analyses of the frontal area length. However, this condidtion is not reflected in the analytic results (below); within-collection variation in the anterior border dimensions is continuous. Some localities within the heterolithic facies are character¬ ized by cranidia with a more strongly developed anterior border furrow than is usually found within this facies. They do not appear to have suffered less flattening than specimens from other localities. Specimens from LaGrange Mountain (RWa) consistently show a marked anterior border furrow, often with associated pits and caeca developed across the frontal area (Hughes, 1993, pi. 4: figs. 6-8, pi. 5: fig. 6; see below). Specimens from Arcadia Bed 18 (AAa; Plate 3: Figures 2-4) NUMBER 79 9 also show an anterior border furrow in all but the largest specimens, in which it may have been obliterated during compaction (Hughes, 1993). Specimens from these localities have relatively long (sag.) frontal areas (the importance of this locality-based variation is discussed below). There are no grounds for considering the anterior border a species-specific character. Despite its variable condition, however, it has long been considered to be of taxonomic significance in Dikeloceph- alus. The development of the anterior border furrow and inflated border in Dikelocephalus minnesotensis was first recognized by Hall (1863), and illustrated with a single fragmented cranidium showing a swollen border (Hall, 1863, pi. 9: fig. 12). This specimen, from the Saukia fauna heterolithic beds at LaGrange Mountain (RWa), has a markedly convex anterior brim with pits weakly developed in the anterior border furrow (Hughes, 1993, pi. 4: figs. 7, 8). Hall referred to this specimen as D. minnesotensis var. limbatus, commenting that “it may be imprudent to multiply specific designations for such remarkable forms as the Dikelocephalus minnesotensis ” (Hall, 1863:141). Walcott (1914, pi. 65: figs. 5-8) illustrated several speci¬ mens that he considered conspecific with D. minnesotensis var. limbatus. He assigned these to Dikelocephalus ? limbatus Hall. On morphological and stratigraphic grounds the association of these specimens seems unlikely and the validity of Walcott’s species is therefore dubious. Walcott’s material is all signifi¬ cantly older than Hall’s specimen. Most of these specimens have since been reassigned to different species or genera. One pygidium (Walcott, 1914, pi. 65: fig. 8), from the Mazomanie Formation, near Baraboo, was transferred to Briscoia winchelli (Resser, 1942). The cranidium and pygidium illustrated by Walcott (1914, pi. 65: figs. 6, 7) from the laminated sand subfacies at Osceola, Wisconsin, belong to the Osceolia osceola Subzone and were assigned to Dikelocephalus th¬ waitesi by Ulrich and Resser (1930). A cranidium (Walcott, 1914, pi. 65: fig. 5) from the Osceolia osceola Subzone heterolithic facies at Osceola is here considered to belong to Osceolia. An inflated border area was used in the diagnoses of three species of Dikelocephalus: D. marginatus, D. norwalkensis, and D. thwaitesi (Ulrich and Resser, 1930). These species, in amended form, were maintained by Raasch (1951). Dikelo¬ cephalus freeburgensis Kurtz (Bell, Feniak, and Kurtz, 1952) from the Reno Member also shows an anterior border furrow (Plate 9: Figures 3-5). Dikelocephalus marginatus came from heterolithic facies at LaGrange Mountain (RWa). This locality has been discussed above. Dikelocephalus thwaitesi and D. norwalkensis came from sandstone facies and are preserved with putative original relief. All other species based on cranidia by Ulrich and Resser (1930) came from heterolithic facies, where compaction has reduced the prominence of the anterior border. As shown above, the development of the anterior border is variable within collections and can not be considered significant in the diagnoses of these species. Some Dikelocephalus specimens from outside the northern Mississippi Valley also show variable development of the anterior border. A specimen that is preserved in original relief from the Snowy Range Formation, Montana (Grant, 1965, pi. 15: fig. 1), clearly shows an anterior border furrow and raised border. Specimens from the Rocky Mountains of Alberta (Westrop, 1986, pi. 3: figs. 1-3) lack an anterior border furrow, although the border of the largest specimen (Westrop, 1986, pi. 3: fig. 2), which is slightly compacted, appears to be slightly inflated. These specimens support the suggestion of locality- based variation in the development of the anterior border. The cranidium from the Upper Franconian of the Arbuckle Mountains, Oklahoma, assigned to Dikelocephalus species 1 by Stitt (1971, pi. 3: fig. 20) also possesses a deep anterior border furrow that widens (sag.) adaxially on its short (sag.) frontal area. In addition to the unusual abaxial widening of the anterior border furrow, the glabella is more elongate than in Dikelocephalus. This specimen is here considered to belong a different genus. Lochman (1959) stated that the absence of an anterior border furrow in Dikelocephalus was a useful character for distin¬ guishing this genus from Briscoia. She also considered the possession of a pair of posterolateral pygidial spines to be characteristic of Dikelocephalus. It is not clear what she regarded as the taxonomic status of described Dikelocephalus species that possess both an anterior border and pygidial spines. Westrop (1986) argued that specimens of Briscoia dalyi (Walcott) show progressive effacement of the anterior border furrow through holaspid ontogeny, making the absence of a border furrow unreliable for use in distinguishing between Briscoia and Dikelocephalus. Westrop’s (1986, pi. 2: figs. 1, 2, 5-8, 10) illustrations show a clear size-related trend from small specimens with firmly impressed anterior border furrows to larger specimens with very shallow furrows; this may reflect ontogenetic control. The variability in the anterior border of Dikelocephalus, on the other hand, appears to be size- independent. Caeca. —Caeca are not prominent in Dikelocephalus. Where obvious, they are usually found in association with a well-developed anterior border furrow. Specimens from La- Grange Mountain, Redwing (RWa), Minnesota, may show the development of irregular caeca, together with pits in the anterior border furrow in small individuals (Hughes, 1993, pi. 4: figs. 6-8, pi. 5: fig. 6). The spacing between pits is highly variable, even within individual cranidia. Similar caeca are seen in specimens from Galesville and Stoddard, Wisconsin, and are about 0.5 to 2.0 mm wide. Caeca are probably present in many other specimens (e.g., Plate 3: Figures 3, 4, Plate 10: Figures 14, 15, Plate 11: Figures 1, 2) but are less distinct and cannot be identified with certainty. Caecal development is variable within collections. The original description of Dikelocephalus minnesotensis var. limbatus (Hall, 1863:141) mentioned “a few wrinkled striae, directed toward the glabella” that originated in the 10 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY anterior border furrow (Hughes, 1993, pi. 4: figs. 6, 7). These structures have also been reported in Briscoia sinclairensis Walcott (see Walcott, 1924, pi. 20: fig. 2) and in Briscoia specimens from central Texas (see Bell and Ellinwood, 1962, pi. 52: fig. 5) but not in any other dikelocephalid. Caeca appear in other libristomate trilobites (see Fortey, 1974; Jell, 1978; Babcock, in press, for examples) and are considered to reflect the morphology of soft parts underneath the dorsal surface. The development of the pits and caeca in Dikelocephalus appears to be much less regular than in most other primitive libristomates (Fortey, 1974; Jell, 1978). The absence of caeca in most specimens does not necessarily imply that the soft parts that they represented were absent in other Dikelocephalus. Eye Ridges. —Faint eye ridges are present in many specimens of Dikelocephalus from all localities studied (Plate 2: Figure 3, Plate 3: Figure 6, Plate 6: Figures 7, 8). They are particularly obvious in larger specimens, in which the palpebral lobes are relatively small, and where the preoccipital fixed cheek is wider (see below). In some specimens they continue into a slight swelling of the interocular fixed cheek opposite the midpoint of the palpebral lobe (Plate 4: Figures 8, 9). The occurrence of the eye ridges is gradational and their morphol¬ ogy is variable within collections. The presence of the eye ridge in D. minnesotensis was discussed by Ulrich and Resser (1930) but was not used as a diagnostic feature in their species designations. Lochman (1959) noted that the eye ridges in Dikelocephalus are faint. Eye Position. —The position of the palpebral lobe varies within Dikelocephalus. In specimens from the St. Lawrence Formation the eye is centered opposite or slightly anterior of the midpoint of the first glabellar lobe (e.g., Plate 3: Figures 1-7, Plate 6: Figures 1-14). Some specimens from the TUnnel City Group have the eye centered opposite, or just posterior of, the confluence of the first glabella furrow with the axial furrow (Plate 9: Figures 1-5, 8-11). Other specimens have it in the same position as in St. Lawrence Formation Dikelocephalus (Plate 9: Figure 15). There is continuous gradation between these two conditions within collections from both the Thnnel City Group and the St. Lawrence Formation, suggesting that eye position is not a species-diagnostic character. The position of the eye lobe has been considered to be of taxonomic importance for distinguishing Briscoia and Dikelo¬ cephalus (Westrop, 1986). In the TUnnel City Group specimens the eye is centered in the same position as in Briscoia septentrionalis Kobayashi (see Palmer, 1968) and Briscoia angustilimba Westrop, 1986. The position of the eye, therefore, does not seem to be genera specific in the dikelocephalids. Ornamentation. —Development of surface ornamentation on the frontal and posterolateral borders is variable. In most cranidia transverse terrace lines are present on the frontal area, just anterior of the glabella (Plate 3: Figures 2-4, Plate 4: Figure 4, Plate 6: Figures 1-14). The occurrence of these terrace lines is independent of size (Plate 6: Figures 1-14). They are continuations of those developed on the free cheeks and tend to become wavy and inosculate towards the ocular region (Plate 6: Figure 7). Some specimens show terrace lines on the posterolateral border, in addition to two transverse striations that are developed adaxially on the spatulate posterolateral border. These features are only preserved in specimens from the fine-grained heterolithic facies, and their absence from speci¬ mens preserved in sandstones is likely due to preservational factors. Glabella Glabellar characters have been considered to be especially important in trilobite taxonomy (Stubblefield, 1936; Fortey and Owens, 1979; Whittington, 1981; McNamara, 1986; Fortey and Chatterton, 1988) and Ulrich and Resser (1930:18) commonly used the shape, contour, and furrows of the glabella in the discrimination of Dikelocephalus species. Some aspects of glabellar shape and ornament are described in this section, whereas others are treated bivariately. Variation in relief of the glabella is commonly the result of compaction (Hughes, 1993), but well-preserved specimens show variation even between specimens from individual localities (Plate 6: Figures 7, 9, 11-14). Anterior Margin. —Well-preserved material shows that the shape of the anterior margin of the glabella varies slightly, from well-rounded to more truncate, among specimens from the same localities (e.g., Plate 4: Figures 4, 5, Plate 6: Figures 2, 6-11, 13). However, this variation is not obviously related to either size or locality, and is not linked to that of any other character. Ulrich and Resser (1930) often commented on the shape of the anterior glabella in their species designations. They considered the axial furrow in front of the glabella in Owen’s (1852) syntype of D. minnesotensis to be “decidedly straight¬ ened in the middle” (Ulrich and Resser, 1930:21) and used this character to distinguish D. minnesotensis from their D. hotchkissi in which “the anterior part of the outline [of the glabella] is more rounded” (Ulrich and Resser, 1930:23). Re-examination of Owen’s original material (Plate 11: Figure 17; Labandeira and Hughes, 1994, pi. 1: fig. 1) shows that the specimen has been subjected to compaction-related distortion, which straightened the anterior margin of the glabella. The orientation of the specimen and retouching of photographs by Ulrich and Resser (1930, pi. 1: fig. 6) also over emphasized the straightness of the margin. Lateral Glabellar Furrows. —All specimens oiDikelo¬ cephalus show at least one pair of glabella furrows (SI). The second pair (S2) is absent only in specimens showing considerable compaction (Plate 3: Figures 5, 6) and was probably present in all specimens. There are three important ways in which the glabella furrows vary within Dikelocepha¬ lus: (1) presence of the third glabellar furrow (S3); (2) NUMBER 79 11 confluence of the first and second glabellar furrows (S1 and S2) with the axial furrow: and (3) angular relationship of SI and S2 furrows to the sagittal axis. S3 is most common in small specimens (Plate 2: Figures 1-3, Plate 3: Figures 1-4, Plate 4: Figure 4, Plate 6: Figures 1-7), but may also be present in large specimens (Plate 6: Figure 11). Owen’s (1852) original, discussed by Ulrich and Resser (1930), shows S3 and is above average size (bl = 4 cm) (Plate 11: Figure 17; Labandeira and Hughes, 1994, pi 1: fig. 1). Many of the largest specimens, however, show only faint S3 furrows and in many cases S3 furrows are absent (Plate 2: Figure 8, Plate 4: Figure 9, Plate 6: Figures 13, 14, Plate 7: Figure 1). The presence of a third pair of glabellar furrows in Owen’s syntype cranidium of Dikelocephalus minnesotensis was noted by Ulrich and Resser (1930:21-22) who observed that “the anterior pair is indicated, sometimes only on one side, by very obscure impressions that appear to turn forward rather than [run] directly across and do not start at the dorsal [i.e., axial] furrow.” Inspite of their recognition of variation in S3, Ulrich and Resser did not place any taxonomic importance on the occurrence of these furrows. Labandeira (1983:164-165) commented that the number of occipital and lateral glabellar furrows is variable, ranging from two in some large specimens to four (or five?) in smaller specimens. Of these the S3 furrows are usually faint, incomplete and highly variable in their angular relationship with the axial furrow. New material confirms Labandeira’s (1983) suggestion that the number of furrows is size-related. However, Labandeira’s interpretation of five cephalic axial furrows in some of the smaller specimens is questionable. A variety of minor intercalated furrows are present in addition to the occipital and glabellar furrows in a few specimens, but these are not considered homologous with the major lateral glabellar furrows and are discussed separately (see below). Within individual collections of Dikelocephalus some specimens show confluence of S1 and S2 with the axial furrow, and others do not (Plate 2: Figures 1-3, 6-8). This pattern of variation is apparently unrelated to size or to locality. The angular relationship of the medial portions of SI, S2, and S3 to the sagittal axis has been examined in detail by Labandeira and Hughes (1994, fig. 2), who concluded that variation in this character was continuous. Analysis using the planvariograph showed no consistent size-related or locality- based variation in this character (e.g., Plate 6: Figures 1-14), thus confirming Labandeira and Hughes’ conclusion. Many specimens show additional furrows, which are termed intercalated furrows (e.g., Hughes, 1993, pi. 5: figs. 6, 7). Although they are sometimes faintly present in material retaining original relief, intercalated furrows are particularly common in slightly compressed specimens (Hughes, 1993, pi. 5: fig. 7). Slight compression may actually enhance these features (see Henningsmoen, 1960, for examples of compres¬ sion-enhancement of internal features). There are no obvious controls on their occurrence other than compression. Minor “furrows” developed within the lateral glabellar lobes are present in a wide variety of trilobites (e.g., the ceratopygid Cermatops discoidalis (Salter): Hughes and Rushton, 1990; the dikelocephalid Briscoia platyfrons Ulrich and Resser: Walcott, 1914, pi. 65: fig. 4). Median Occipital Tubercle. —A small tubercle is present on the occipital lobe of just more than half the Dikelocephalus cranidia studied from four localities in the heterolithic facies (Ferguson’s collection from Stillwater (SWb), Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8)) (Figure 4). The occurrence of the median tubercle is not size-related, as it is present in approximately half the samples in each cranidial size class (Figure 5). The tubercle apparently maintains the same size relative to the glabellar length throughout growth (Plate 2: Figures 1-3, 6, 7, Plate 3: Figures 1-5, Plate 6: Figures 1-14, Plate 10: Figures 14, 15). Interpretation of the occurrence of the median tubercle is complicated by the tubercle’s morphological variability. In some specimens it is prominent (Plate 6: Figures 8, 11); in others it has low relief and is difficult to distinguish from glabellar pustulation (Plate 6: Figure 3). The absence of the tubercle is not due to taphonomic factors because some well-preserved specimens lack any indication of a median tubercle (Plate 6: Figures 4, 9). As the median tubercle is present in approximately half the specimens and is size- independent, it may represent a dimorphic character. Dimor¬ phism in sclerite characters has been reported in many trilobites (e.g., Selwood and Burton, 1969; Hu, 1971; Campbell, 1977) and is commonly sexually-related in modem arthropods (Hartnoll, 1982). However, the prominence of the tubercle appears to vary continuously and this variation is not linked to that of any other character. Ulrich and Resser’s (1930:33) discussion of D. oweni includes reference to “the usual median tubercle.” It seems that Median tubercle: □ present ^ absent n = 122 Figure 4.—Pie diagram showing percentage occurrence of median tubercle in Dikelocephalus cranidia from four localities in the St. Lawrence Formation. Localities are Stillwater (SWb), Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). 12 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY a> o c a> o o o >* o c a> 3 O" — 3 o o o >* o c Pustules: □ present H absent glabellar size class: bl (cm) FIGURE 6. —Bar chart showing relationship between occurrence of glabellar pustulation and size. Arrows mark the size class of the largest pustulated cranidium of Dikelocephalus at each of four localities in the St. Lawrence Formation; n = 147. Localities are Stillwater (SWb), Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). they thought it was present in all Dikelocephalus, as this feature was not discussed elsewhere in their monograph. ORNAMENTATION.— Most small cranidia of Dikelocephalus have pustules developed on the glabella; all large cranidia lack them. The restriction of pustules to smaller forms is unlikely to be a taphonomic artifact, because many large cranidia that lack pustules show excellent preservation of other surface ornament (Plate 6; Figure 11, 13). Populations from single bedding planes show that pustulation is absent in all specimens with an occipital glabellar length greater than 3.5 cm (Figure 6). The NUMBER 79 13 absence of pustules in some small individuals may be due to taphonomic factors. In very small cranidia the pustules are so tiny that the sediment grain size of even the fine-grained units is too coarse to permit their preservation. All small specimens that show surface ornamentation on other parts of the cranidium also show pustules. It is therefore likely that all small Dikelocephalus from the northern Mississippi Valley were pustulated in life. Examination of collections from single bedding planes reveals no clear division between the pustulated and non- pustulated forms. Pustules in Dikelocephalus are distinctive in small cranidia but in larger specimens they are less densely concentrated and of lower relief (Plate 6: Figures 1-10). This transition forms a gradual and continuous trend toward total loss of pustules at large size. As glabellar size increases pustules become restricted to axial regions (Plate 6: Figures 9, 10). These observations indicate that pustule development was ontogenetically controlled and that all individuals eventually lost their pustules during ontogeny. There is no evidence of separate pustulated and non-pustulated morphs (as are reported in Dechenella, by Selwood, 1965, or Paciphacops logani (Hall), by Campbell, 1977). The size of cranidia at which the last remnant of pustulation is detectable is approximately the same in all localities (Figure 3), suggesting that this control was locality-independent. A similar relationship was present throughout all the collections examined from the heterolithic facies. Although taphonomic factors prevent assessment of the maximum size of individuals living at each locality, it seems probable that the size range of Dikelocephalus was locally variable, as is common in other trilobite species (Sheldon, 1988). The observation that pustules become obsolete at approximately the same size at all localities suggests that there may have been an adaptive reason for pustule loss at a particular size. To explore the growth parameters of pustulation the diameters of five pustules were measured on 34 cranidia from Stillwater (SWb), Arcadia (AAa), and North Freedom (NF2, NF8). These values were averaged and then plotted against occipital-glabellar length (bl). Results show a strong correla¬ tion between these variables (r = 0.880) (Figure 7). The reduced major axis coefficient (0.883) is not significantly different from isometry at the 95% confidence level, but a plot of the ratio of pustule diameter to glabellar length against occipital-glabellar length (Figure 8) suggests a slight decrease in the relative size of the pustules in larger cranidia. Most small cranidia from outside the northern Mississippi Valley, preserved in fine-grained limestones, do not have pustules (Plate 11: Figures 11, 15; Grant, 1965, pi. 13: fig. 1; Westrop, 1986, pi. 3: fig. 1). All surface ornament is lacking in these specimens and the significance of the absence of pustulation is difficult to evaluate, although other trilobite species from the same lithologies display surface ornament. A single cranidium preserved in limestone from Hudson Creek, Texas (bl = 0.8 cm, Plate 11: Figure 14), does show small pustules, indicating that at least some pustulated individuals lived outside the northern Mississippi Valley. Hence biological variation in Dikelocephalus ornament occurs throughout its geographic range. Glabellar ornamentation was used by Ulrich and Resser (1930) to designate two small species, D. granosus and D. harretti. The results above indicate that pustulation is not an appropriate character for specific designation in Dikelocepha¬ lus and these species are therefore considered to be invalid. The fact that few specimens illustrated by Ulrich and Resser (1930) show pustulation reflects their bias toward selecting larger specimens for illustration, in which pustulation is consistently absent. Raasch (1951) suggested that pustulation was common among small cranidia of Dikelocephalus in his Saukia sublonga faunal unit of the heterolithic facies and reported that pustulation persisted into medium-sized forms in the underly¬ ing Saukia subrecta unit. Raasch (1951) correctly implied that pustulation was ontogenetically controlled but thought that its development varied through time. The results above do not support the conclusion that there was stratigraphic control of pustule development. Collections analyzed here came from throughout the heterolithic facies, but all show pustule loss at a similar size. Labandeira (1983) noted that pustulation was not present on all small specimens, and suggested that pustule variation should be classed as ecophenotypic rather than ontogenetic in origin, although the data were ambiguous. Labandeira’s observation was based on Ulrich and Resser’s material which included disproportionately few small specimens. This fact, coupled with the abraded condition of much of this material, lead to under-representation of pustule occurrence in the type suite. Free Cheeks Median suture. —A striking morphological variation wi¬ thin Dikelocephalus is the intermittent presence of the median connective suture, which joins the dorsal facial suture with the hypostomal suture. The median suture was considered synapo- morphic for the Order Asaphida (Fortey, 1990) by Fortey and Chatterton (1988), who demonstrated loss of the structure in several groups (see below). Complete cephala of Dikelocepha¬ lus possessing a median suture are rare. Ulrich and Resser (1930, pi. 10: fig. 2) figured a specimen in which the original orientation of the free cheeks had been disturbed. In this specimen a median suture is present (Labandeira and Hughes, 1994, pi. 1: fig. 8). A median suture may be present in an extremely large pair of free cheeks from Muscoda, Wisconsin (Hughes, 1993, pi. 4: fig. 4). However, this structure could be a crack, as its course is slightly irregular. Many isolated free cheeks show a sharp truncation of the doublure along the sagittal axis, which is likely to represent a median suture (Plate 14 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY occipital-glabellar length: bl (mm) FIGURE 7. —Bivariate scatterplot showing relationship between average pustule size (mean of five pustules measured on each cranidium) and occipital-glabellar length in Dikelocephalus from four localities in the St Lawrence Formation; n = 34. Localities are Stillwater (SWb), Arcadia Bed 18 (AAa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). FIGURE 8.—Bivariate scatterplot of ratio of average pustule size divided by occipital-glabellar length plotted against occipital-glabellar length for same data set as in Figure 7. Note the general decrease in relative pustule size. 3: Figure 7, Plate 5: Figure 1, Plate 7: Figure 5; Labandeira and Hughes, 1994, pi. 1: fig. 10). This suggests that a functional median suture was commonly present in Dikelocephalus. However, there are also many specimens of fused free cheeks that lack evidence of a median suture (Plate 1: Figure 2, Plate 2: Figure 4, Plate 3: Figure 8, Plate 5: Figure 2). In these cases the doublure is continuous across the sagittal axis. There is little evidence of fused (i.e., non-functional) median sutures in Dikelocephalus. A lineation, which has not developed com¬ pletely across the doublure, is present in one yoked pair of free cheeks from North Freedom Bed 2 (Hughes, 1993, pi. 4: fig. 2). This is slightly irregular and may be a crack, rather than a partially fused median suture. There is no obvious control on the presence or absence of the median suture. All collections studied from the St. Lawrence Formation include specimens that have and lack it. Its occurrence is not size-related (Plate 5: Figures 1, 2; Hughes, 1993, pi. 4: fig. 2). Ulrich and Resser (1930:42) suggested that specimens that possessed a median suture showed a facial suture that was “entirely dorsal, passing around the brim of the cephalon ... inside the [anterior] edge of the cephalon.” New material disproves this. Some specimens, which clearly lacked NUMBER 79 15 a median suture, had a marginal facial suture (Plate 2: Figure 4, Plate 2: Figure 8; Hughes, 1993, pi. 4: fig. 2). Others, which had the facial suture retracted, (see above) probably possessed a median suture (Plate 5: Figure 1). In fact, variation in the presence of the median suture is not related to variation in any other character in Dikelocephalus (Rasetti, 1952). Because there is no obvious pattern of occurrence, and because it is not obviously linked to variation in any other character, the presence of the median suture is not considered species diagnostic. It was not essential in the molting process of Dikelocephalus (see Hughes, 1993) and in some cases it appears to be only partially developed. Conjoined free cheeks are also present in Dikelocephalus from the Tlinnel City Group. The presence or absence of the median suture was used by Ulrich and Resser (1930) as a key character for distinguishing two groups within the genus: the oweni group, including D. oweni and D. raaschi, which possessed a median suture; and the gracilis group including D. retrorsus, D. ovatus, D. granosus, D. gracilis, and D. subplanus, which lacked the suture. Rasetti (1952:892) questioned the validity of a large number of Dikelocephalus species from the northern Missis¬ sippi Valley noting that “in other species of the genus, almost indistinguishable from [Dikelocephalus retrorsus and D. subplanus ] in other characters, there is a median suture.” The results of this analysis support Rasetti’s (1952) view; no taxonomic significance can be attributed to the variable nature of the median suture. The interpretation of variation in the presence of the median suture as an intraspecific variable is consistent with the pattern shown in Isotelus gigas DeKay (Henningsmoen, 1975; Jaanus- son, 1975; Ludvigsen, 1979) and is also characteristic of other asaphide species (Ludvigsen, 1979). The loss of the median suture in some kainellids by the fusion of the free cheeks is recognized as a secondary condition in the Remopleuridacea (Fortey and Chatterton, 1988). It appears that fusion of the free cheeks was a common secondary condition in asaphide trilobites. Lateral Border.—A slight marginal swelling is present in many specimens of Dikelocephalus (Plate 3: Figure 7, Plate 7: Figure 7) and is often associated with a weakly impressed lateral border furrow and terrace ridges (Plate 2: Figures 10, 11). It marks the lateral extension of the anterior border. Even in specimens preserved with original relief it is never prominent (Plate 9: Figure 6; Hughes, 1993, pi. 1: fig. 2, pi. 5: figs. 4, 5). Variation in the marginal border of Dikelocephalus is slight and probably compaction-related. Ulrich and Resser (1930:18) stated that the presence of the marginal border (and associated lateral border furrow) was useful in the distinction of species, particularly in D. norwalkensis, which is preserved with its original relief in laminated sandstones, but also in D. marginatus from the heterolithic facies. A reexamination of this material (Labandeira and Hughes, 1994, pi. 1: fig. 10) suggested that a slightly raised rim is present in the syntypes of D. marginatus but that it is no more prominent in these specimens than in others of Ulrich and Resser’s species (e.g., D. declivis; Ulrich and Resser, 1930, pi. 15: fig. 9). Length of Genal Spine. —The length of the genal spine is variable in Dikelocephalus (Plate 1: Figures 1, 2, Plate 5: Figure 1, Plate 7: Figure 5). It probably extended at least as far as the anterior of the pygidium in smaller specimens (Plate 1: Figures 1, 2). The entire length of the spine is rarely preserved so it has not been possible to draw conclusions on the nature of its variation. Ulrich and Resser (1930) based some of their species designations on the inferred nature of the termination of the spine as interpreted from the morphology of its base. In the description of D. ovatus they stated that “the genal spines themselves are not completely preserved but what remains of them shows that they have broader bases and suggests that they are also shorter than in D. gracilis ” (Ulrich and Resser 1930:45). In contrast, they suggested that D. wisconsinensis was characterized by the “great length of the genal spines” (Ulrich and Resser 1930:41). The present examination suggests that the supposed inverse relationship between the width of the spine at its base and its total length cannot be established conclusively. Although Labandeira (1983:139) did not include genal spine length in his morphometric analysis of Ulrich and Resser’s type suite, he suggested that genal spines are more morphologically variable than glabellae. Ornamentation. —The number and form of dorsal terraces varies slightly between specimens within localities but is not size-related. Ridges tend to become inosculate and granular on the ocular platform. One specimen, preserved with original cuticle, shows granulations that are hexagonal (Hughes, 1993, pi. 1: fig. 2). The unique occurrence of such structures is attributed to the unusual preservation of the cuticle and is not considered taxonomically important. Hypostome (Labrum) The planvariograph method (see above) indicated that measurement of the hypostome was unlikely to provide information on growth controls within Dikelocephalus; mor¬ phometric analysis of the hypostome was restricted to measurements relating to the size of the maculae. Analysis of the type suite showed that the relative length and width of the hypostome remained constant through ontogeny (Labandeira and Hughes, 1994, fig. 8). The median body shows some variation in convexity, but this is likely to be influenced partially by compaction. Large specimens frequently show cracks developed along the sagittal axis. The length/width ratio of the median body remained constant through growth but well-preserved specimens suggest that the median body may have become less convex during ontogeny (Plate 2: Figures 5, 9, Plate 7: Figures 9-12), supporting a similar observation made by Labandeira (1983). The fine “thumb print” terracing evident on the median body 16 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY of many large hypostomes (Plate 3: Figure 9; Hughes, 1993, pi. 7: figs. 1, 2) was probably originally present in all specimens. Maculae are more common in small specimens than in large ones (Figure 9; Labandeira and Hughes, 1994, pi. 1: fig. 11). Hypostomes from individual bedding plane collections show a trend toward effacement of maculae at large sizes (Plate 2: Figures 5, 9, Plate 5: Figures 3, 4) but sample sizes are too small be assessed statistically. As the largest specimens are often excellently preserved (Plate 7: Figure 11; Hughes, 1993, pi. 7: figs. 1, 2) the progressive effacement of maculae is not merely due to compaction. Not all small specimens show well-developed maculae (Figure 9), possibly indicating that the pattern of variation was not controlled by size alone. Investigation of possible locality-related influences on maculae development is prohibited by small sample sizes (see Hughes, 1993). Maculae apparently grew at a rate directly proportional to glabellar size, but became gradually effaced (i.e., less prominent) through holaspid ontogeny. Ulrich and Resser (1930) were reluctant to base species designations on the basis of hypostomal characters alone, although they thought that the prominence of maculae varied between species. Labandeira (1983) and Labandeira and Hughes (1994) considered macular variation to be “intraspeci- fic,” a conclusion supported by this study. Thorax Number of Thoracic Segments. —The smallest and largest articulated specimens show nine thoracic segments, but their thoraxes are not complete (Plate 1: Figures 1, 2). In the largest specimen the pygidium is telescoped over part of the thorax (Plate 1: Figure 1) and it is not possible to determine whether other segments are hidden beneath it. The smallest specimen is rather poorly preserved and lacks the cranidium (Plate 1: Figure 2). An inverted segment lies within the cephalic body cavity and originally may have been part of this individual, lying at the anterior of the thorax. There were probably at least ten thoracic segments during life. In both these specimens the pygidium lies slightly oblique to the cephalon, suggesting that they may represent molts rather than dead individuals (see Henningsmoen, 1975). A third specimen has twelve segments, and although the cephalon and pygidium are poorly preserved, the thorax appears to be complete (Plate 1: Figure 3). Some, if not all, Dikelocephalus had twelve thoracic segments in the holaspid stage. The difference in segment numbers between the three specimens could be accounted for by taphonomic factors alone. However, it seems unlikely that as many as three segments are missing from the largest specimen (Plate 1: Figure 1) and there may be additional biological reasons for the variation in segment number. The variation is unlikely to be growth-related, as the specimen showing the smallest number of segments is the largest of the three. Intraspecific variation in the number of thoracic segments is not uncommon in trilobites. The thorax of adult Paradoxides davidis brevispinus Bergstrom and Levi-Setti contains 19 to 21 segments and other subspecies show similar variation in segment number (Bergstrom and Levi-Setti, 1978). Within the ptychoparioid species Elrathia kingii (Meek) the adult thorax contains 10 to 13 segments (Bright, 1959). a> o e a> o o o u c a> 3 cr a> Maculae: □ present H absent o in 1—pill CD CT> , p O) CT> O) 0> *7 *7 c\i C\i CO o 6 6 6 o o in o in o c\j OJ CO macular sizes classes: hi (cm) FIGURE 9.—Bar chart showing relationship between occurrence of maculae and the length of the hypostome in Dikelocephalus from four localities in the St. Lawrence Formation; n = 39. Localities are Stillwater (SWb), Arcadia Bed 18 (AAa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). NUMBER 79 17 Anterior Margin of Thoracic Segments. —Study of new and more complete material shows that the shape and furrows of the thoracic segments varied markedly along the thorax (Plate 1: Figures 1-3). Ulrich and Resser (1930) considered the characteristics of the thoracic segments to be important for identifying D. orbiculatus. The anterior margin of an isolated segment attributed to D. orbiculatus shows a uniform curve (Ulrich and Resser, 1930, pi. 4: fig. 4), which they considered to be diagnostic of this species. However, most segments near the posterior of the thorax show this uniform curve and, as it is impossible to determine the position of this isolated segment in the thorax. The use of curvature in taxonomic discrimination cannot be justified. Ornamentation. —The petaloid ornamentation of the ante¬ rior facets of thoracic pleurae varies in appearance, probably due to preservational control. A faint median tubercle is evident on the axial ring of a small number of specimens (Plate 2: Figure 12). Granular pustules are developed on adaxial portions of the opisthopleurae and on the axial ring of some smaller segments (Plate 3: Figures 10, 11, Plate 5: Figure 5, Plate 7: Figure 13). These pustules are absent in larger specimens (Hughes, 1993, pi. 4: fig. 5), probably reflecting an ontogenetic control. Pygidium Axial Furrows. —Four axial rings and a terminal piece are present in all Dikelocephalus except those showing teratologies (e.g., Hughes, 1993, pi. 7: fig. 7). The furrow defining the posterior margin of the fourth ring is variably incised but prominent (Plate 5: Figures 6, 8, 10, 11). There are no obvious size or locality controls on its prominence. A few specimens show up to two weak and incomplete furrows on the terminal piece (Plate 5: Figure 11; Ulrich and Resser, 1930; Lochman, 1959). They are similar in appearance to those described in species of Centropleura (Opik, 1961, pi. 8; Babcock, in press, figs. 18, 20). These features are too rare to permit any assessment of likely controls on their occurrence. An anteriorly facing ridge arches from the posterior border of the first two axial rings in many specimens (Plate 5: Figures 9, 10, Plate 8: Figure 8). This structure is probably an incompletely developed articulating half ring, as in Ceraurus (Whittington, 1959). Termination of the Axis. —The shape of the terminal piece is somewhat variable within Dikelocephalus. In this analysis it was assigned to one of three character states: smooth (e.g., Plate 8: Figures 11, 12), intermediate (e.g., Plate 8: Figures 5, 6), or sharp (e.g., Plate 8: Figure 1). The boundaries between these states were arbitrary, reflecting a morphological continuum. Results suggest that termination states are ran¬ domly distributed both with respect to locality and to pygidial size (Figure 10). There may be a slight overall trend toward a smooth termination in larger specimens. Ulrich and Resser (1930) used the shape of axial termination in the designation of several species. They suggested, for example, that D. orbiculatus had a “semi-elliptical bulblike termination rather than the slender obconical one seen in D. hotchkissi" (Ulrich and Resser, 1930:30). This analysis shows that there is no justification for using this character in specific designations. Post-axial Ridge. —The axis appears to extend into a low and poorly-defined post-axial ridge in a small number of specimens (Labandeira and Hughes, 1994, pi. 1: fig. 14). Uncompressed material shows that even where this ridge is not developed, the dorsal surface arches upward slightly posterior of the axis (Hughes, 1993, pi. 6: fig. 6). Ulrich and Resser (1930) suggested that D. norwalkensis (only found in the laminated sand subfacies) was characterized by the possession of a post-axial ridge. The prominence of this ridge was over-emphasised in the retouched photographs (e.g., Ulrich and Resser, 1930, pi. 21: fig. 16) and it is not a species-specific character among Dikelocephalus. Post-axial Emargination. —The development of a slight post-axial emargination is common in Dikelocephalus from all localities (Plate 4: Figures 2, 3, Plate 5: Figure 12, Plate 11: Figures 3, 5). It probably represents an upward arch of the dorsal surface posterior of the axis, rather than a marked inflection of the margin (Plate 11: Figures 8, 9). A similar emargination has been recognized in Cermatops discoidalis (Salter) (Hughes and Rushton, 1990), where its prominence may depend on the orientation of the pygidium during preservation. Emargination appears to be present in many transversely macropygous pygidia. The post-axial emargina¬ tion is present in large specimens of Dikelocephalus and the degree to which it is developed may be slightly size-influenced (Figure 11). Ulrich and Resser (1930:39) noted that in some specimens the “posterior edge ... is gently sinuate” and considered this feature to be diagnostic of Dikelocephalus raaschi. Twenhofel (1945, pi. 88: fig. 3), found an emarginate specimen that differed from D. raaschi in other characters. In assigning this specimen to D. oweni, he commented that the validity of many of Ulrich and Resser’s (1930) species was doubtful. The results above indicate that emargination is a variable character and is indicative of intraspecific variation. Number of Pleural and Interpleural Furrows. —The combined number of pleural and interpleural furrows on the pygidium of Dikelocephalus varies between 7 and 10; most specimens show either 8 or 9 furrows. In some specimens the number of furrows on each side of the sagittal axis is asymmetrical. The score for these pygidia was calculated as the average between the two sides (as in Sheldon, 1987). The number of furrows is not obviously size or locality-related (Figure 12). There is no suggestion that the development of incipient axial rings beyond the fourth ring is related to an increased number of pleural and interpleural furrows. Hence, there is no clear relationship between the number of pleural furrows and the axial furrows of the pygidium. Ulrich and Resser (1930) suggested that 9 to 10 ribs are 18 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 10.—Plot showing relationship between the shape of the axial termination and the length of the pygidium in Dikelocephalus from three localities in the St Lawrence Formation; n = 47. Localities are LaGrange Mountain (RWa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). FIGURE 11.—Plot showing occurrence of a post-axial emargination in relationship to the length of the pygidium in Dikelocephalus from four localities in the St. Lawrence Formation; n = 62. Localities are LaGrange Mountain (RWa). Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). present in Dikelocephalus but did not use this variation in designating species. Division of the Pleurae. —Variability in pleural division is a striking feature within Dikelocephalus. Unequal division reflects an enlarged propleural band. Pygidia from four localities within the heterolithic facies have been assigned to one of three character states: equal (e.g., Plate 3: Figure 12); subequal (e.g., Plate 3: Figures 15, 16); or unequal division of pleurae (e.g., Plate 3: Figure 14) (Figure 13a). Analysis indicates that, within these localities, the wide variation in the division of the pleurae is not size-related. Subequal division is the most common condition in specimens from all localities. Slight variation may occur between localities (Figure 13b,C) but the pattern of pleural division is highly plastic. Hence the erection of species solely on this basis pleural division is unwarranted. Dikelocephalus from the Tbnnel City Group often show unequally divided pleurae with an enlarged and inflated NUMBER 79 19 localities: • RWa x NF2 + NF8 pygidial length: zl (cm) FIGURE 12.—Plot showing relationship between number of pygidial furrows and length of the pygidium in Dikelocephalus from three localities in the St. Lawrence Formation; n = 52. Localities are LaGrange Mountain (RWa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). Pygidial furrow number is the total number of pleural and interpleural furrows on the pygidium divided by 2. propleural band (Plate 10: Figures 1-6, 13), but this condition is variable among specimens from the group (Labandeira and Hughes, 1994, pi. 1: figs. 12, 13). Similarly, within the laminated sand facies the pleural division may vary markedly (Ulrich and Resser, 1930; pi. 19: figs. 2, 4, 5, pi. 21: figs. 16-21). Material from the carbonate platform shows similar variability in pleural division (Plate 11: Figures 12, 13, 16). There is no reason to suggest that division of the pleurae is related to sedimentary facies, nor to particular localities. Raasch (1951) suggested that pleural division was size- related in Dikelocephalus. He commented that within the Saukia sublonga faunal unit (equivalent to the heterolithic facies of the Saukia fauna of Hughes, 1990) larger forms were characterized by unequal pleural division. The results of the present study do not support size control of pleural division. Even within localities there is no clear relationship between size and pleural division. Winchell (1874:188) noted that the pleurae of D. minne- sotensis were nearly equally divided “in their outer extension.” Ulrich and Resser (1930) made frequent use of the variable pleural division in their species designations (Table 1). Forms such as D. oweni, D. raaschi, and D. inaequalis were supposedly characterized by unequal division. Labandeira (1983) measured the angle of divergence of the first three pleural furrows in Ulrich and Resser’s type suite of specimens pooled from over 50 localities. His results indicated that the angle of divergence of all three furrows was variable, but a dsitinction of two or more morphs was not apparent (Laban¬ deira and Hughes, 1994, fig. 4). Westrop (1986) referred specimens in a collection from Alberta to D. oweni and illustrated a tiny pygidium showing subequal pleural division (Plate 11: Figure 12; Westrop, 1986, pi. 3: fig. 4). He commented that a larger specimen (Plate 11: Figure 13) shows unequal division and short posterolateral spines, which were supposedly characteristic of this species. Westrop (1986:28) suggested that “subequally divided” pleurae were diagnostic of the genus. The present study suggests that pleural division is highly variable and that no single character state can be used to diagnose species of Dikelocephalus. Where division is unequal in Dikelocephalus, it is the propleurae that are expanded relative to the opisthopleurae. This contrasts with the condition in the rest of the superfamily Dikelocephalacea (in the sense of Ludvigsen and Westrop, 1983a). In these taxa the opisthopleurae are swollen where division is unequal (e.g., Briscoia septentrionalis Kobayashi and “ Briscoia ” elegans (Kobayashi) Palmer, 1968, pi. 15: figs. 3, 4, 6, 7), although Ludvigsen and Westrop mistakenly suggested that propleurae of Prosaukia hartii (Walcott) expanded during ontogeny (Ludvigsen and Westrop, 1983a, text-fig. 6). Their plates (Ludvigsen and Westrop, 1983a, pi. 10: figs. 5, 6, pi. 11: fig. 12, pi. 12: figs. 2, 3) show that there was expansion in the opisthopleurae throughout ontogeny. Length of the Pleurae. —In their description and illustra¬ tion of D. marginatus, Ulrich and Resser (1930) suggested that the first two pairs of pleural and interpleural furrows extended as far as, and converged with, the base of the posterolateral spine. Their retouched illustrations over-emphasised both the extent and the degree of convergence of these furrows. Possible convergence of the furrows is present in a few specimens (Plate 10: Figures 1-6, Plate 11: Figures 12, 16) but is not sufficiently widespread for this relationship to have been convincingly demonstrated. If furrows do converge at the base of the spine, this suggests that the spine is associated with at least two pygidial segments and is not homologous with marginal spines division subequal SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY pygidial length: zl (cm) unequal subequal X X )0< x m XX X X XXX X XX X XX XX X X X X X pygidial length: zl (cm) pygidial length: zl (cm) NUMBER 79 21 FIGURE 13.—Plots showing relationship between division of the pleurae and pygidial length in Dikelocephalus from the St Lawrence Formation; n = 91. A, Specimens are from four populations including LaGrange Mountain (RWa), Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). B, Specimens from North Freedom Bed 2 only. C, Specimens from North Freedom Bed 8 only. in Parabriscoia (Kobayashi, 1935), Elkia (Walcott, 1924a), and Briscoia septentrionalis (Palmer, 1968), which are direct extensions of individual opisthopleurae. The spines of the small pygidium from Alberta (Plate 11: Figure 12) do not appear to be related to the first pygidial segment. If this specimen is a late meraspid, which is likely given its small size, the anteriormost pygidial segment may be homologous with the last thoracic segment in adult holaspid Dikelocephalus. Marginal Rim. —A large specimen from Galesville, Wis¬ consin (Twenhofel, 1945, pi. 88: fig. 3), has a narrow tubular rim developed on the margin of the doublure (Plate 11: Figures 3-5). This structure does not extend along the posterolateral spine. The rim has not been recorded in any other specimen, even of comparable size. As this pygidium is typical of Dikelocephalus minnesotensis in all other respects, the unique occurrence of this minor structure provides insufficient grounds for specific differentiation. The pygidial border of D. marginatus was described as “rimmed” by Ulrich and Resser (1930:50). The nature of this rim was not discussed and re-examination of the syntypes does not suggest a rim on the pygidial margin. Ornamentation. —Pustulation on the pygidial axis is size-dependent and varies in a similar way to glabellar pustulation. Pustules are common in smaller specimens (Plate 3: Figures 14, 15, Plate 5: Figure 6, Plate 8: Figures 1, 2, 5, 6) and are not found in specimens with pygidial lengths (zl) greater than about 5 cm. The pattern of pygidial ornamentation is similar to that on the glabella and suggests ontogenetic control of pustulation throughout the entire axis. The ornament of other areas of the pygidium is variable. A small fragmented pygidium from the laminated sand subfacies near Mauston, Wisconsin, shows pustules on both the propleurae and opisthopleurae (Hughes, 1993, pi. 6: fig. 4). A medium-sized pygidium, from Hokah, Minnesota, shows tiny pustules developed only at the posterior margins of the pleurae (Hughes, 1993, pi. 6: fig. 3). As surface ornament is variable within collections, the unusual condition of these two speci¬ mens is not considered to be of taxonomic importance. The dorsal surface of most specimens is covered with inosculating terraces (more regular transverse ridges are present on the anterior facet of the first pleurae), which become crescentic or granular at the posterior of the terminal piece and on the pleural platform (Plate 5: Figures 6-12, Plate 8: Figures 2-8). There appears to be a gradational transformation from terraces into pustules towards the axis. The terracing on the ventral surface of Dikelocephalus is much more regular than that on the dorsal surface. The distances between four or five terraces in the middle section of the doublure were combined to produce an average terrace spacing value. The analysis included specimens from both heterolithic and laminated sand facies. Correlation between terrace spacing and pygidial length is positive and strong (r = 0.974). Reduced major axis yields a regression coefficient of 1.059, strongly suggesting isometric increase in terrace spacing through growth, even though the sample size is small (n = 18). Hence, terrace spacing increases in direct proportion to pygidial length (Figure 14). and there is no evidence of allometry in terrace spacing during growth. There is little apparent locality FIGURE 14.—Bivariate scatterplot showing relationship of average distance between terraced ridges on pygidial doublure (mean of five measurements) to pygidial length in Dikelocephalus from five localities in the northern Mississippi Valley; n = 18. Localities are Hell Hollow (HW), Lansing (LSb), Hickory Flat Road Quarey, Muscoda (MCa), North Freedom Bed 2 (NF2), and North Freedom Bed 8 (NF8). Hell Hollow and Lansing collections are from the Ttinnel City Group. 22 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY TABLE 3. —Variation in nominal and ordinal characters in Dikelocephalus. "temporal variation implies a difference between specimens from the "Rinnel City Group and the St. Lawrence Formation. Intercollection variation implies a clinal pattern of variation between localities. ? means too few data are available to establish this trend with confidence. Character Variation pattern Comments Ontogenetic Intrapopulational Interpopulational Cranidia Anterior border X X Associated variation in caeca and suture Caeca X X Assocated variation in anterior border Eye position X X temporal variation; similar variation in specimens from outside the region Eye ridge X Glabellar anterior margin X S3 furrows ? X Intercalated furrows X Confluence of SI and S2 with ? X axial furrow Median tubercle X Possibly dimorphic Glabellar pustulation X X Similar variation in specimens from outside the region Free Cheeks Median suture X Possibly dimorphic Marginal rim X Genal spines ? Ornament X Hypostomes Median body 7 X Maculae X X Thorax Thoracic segment number ? Ornament X Pygidium Number of axial furrows X "termination of axis ? X Post-axial ridge X Post-axial emargination 7 X Pygidial rib number X Pleural division X X Possible temporal variation Ventral rim X Dorsal ornament X X Ventral terracing X or substrate influence on this character, as specimens from sandy facies of the Tlinnel City Group fall within the same trend as specimens from the heterolithic facies (Figure 14). DISCUSSION Analysis of twenty-seven nominal and ordinal characters reveals considerable variability within northern Mississippi Valley Dikelocephalus (Table 3; see also Hughes, 1991, fig. 2). Seventeen of these characters were used by Ulrich and Resser (1930) in their species designations (Table 1), five of them were discussed by Labandeira (1983). This analyses shows that these characters provide no grounds for the recognition of more than a single morphospecies. The lack of discontinuous variation within Dikelocephalus suggests that all specimens are part of a single morphospecies. This argument is strengthened by that fact that variations in nominal characters are almost all present within collections from single beds. Several nominal and ordinal characters in Dikelocephalus show evidence of ontogenetic control (Table 2). Pustulation on the axis and maculae development on the hypostome become less prominent as overall size increases. Size-dependent variation in these characters is interesting because it generally is thought that trilobite morphology varies little during the holaspid stage (Shaw, 1956; Whittington, 1957, 1959; Saul, 1967; Pabian and Fagerstrom, 1968, 1972; Hu, 1971; Eldredge, 1972). All the characters of Dikelocephalus discussed above show NUMBER 79 23 variation within bedding plane collections (Hughes, 1991, fig. 2). Most show as much variation within, as they do between, collections. The few characters that do show locality-based differences also show significant intracollection variation. This suggests that intercollection differences are simply exaggerated intracollection variations. Within collections all major charac¬ ters seem to show continuous variation. Even the two putative presence/absence characters (median tubercle and median suture) show gradational development and do not provide strong support for the presence of discrete morphs. There is limited evidence for temporal morphological changes in Dikelocephalus. The position of the eye appears constant within the St. Lawrence Formation, but varies both within and between collections in the TUnnel City Group. It is the only nominal and ordinal character that differs between the Tlinnel City Group and St. Lawrence Formation collections. All Dikelocephalus from outside the northern Mississippi Valley lie within the range of variation present in the region. They also show some of the same intracollection variation. The variation of a few characters on the frontal area may be linked, but the vast majority of characters vary independently. The marked intraspecific variation in nominal and ordinal characters in Dikelocephalus has implications for studies of other trilobites. For example, surface ornament has long been used as a taxonomically significant character in trilobites. It has been suggested that ornament is a “primary species characteris¬ tic” of Upper Cambrian trilobites (Palmer, 1960:53), and the presence of pustulation has been used to define many trilobite species (e.g., Ulrich and Resser, 1933; Chatterton and Perry, 1983; Eldredge, 1972, 1973). Analysis of Dikelocephalus shows that pustulation may show marked variation during the holaspid ontogeny of a single species, which falsifies the suggestion that ornament variation is necessarily species- specific (see also Stubblefield, 1926; Rasetti, 1948:9; Bright, 1959; Ludvigsen and Westrop, 1983a; Fortey and Owens, 1991). Although phenotypic variation in the pustules has been recorded (e.g., Phacops : Eldredge, 1972, 1973; Dalmanites myops (Konig): Ramskold, 1985; Whittington and Evitt, 1953:31), little attention has been paid to the possibility of ontogenetic control of pustulation development. The recogni¬ tion of ontogenetic regulation of pustule development within the holaspid stage of Dikelocephalus was dependent on detailed analysis of a large number of specimens. Pustulation may only be used to define species in cases where the ontogeny of pustules is well known. This investigation of D. minnesotensis shows that many other characters previously considered “species-specific” actu¬ ally represent intraspecific variation. Marked intraspecific variation in nominal and ordinal characters has also been documented in the holaspids of a variety of trilobite clades, including eodiscids (Jell, 1975, 1990); redlichides (Bergstrom and Levi-Setti, 1978); corynexochides (Rasetti, 1948; Robison, 1967); ptychoparioids (Bright, 1959); asaphides (Fortey and Shergold, 1984; Sheldon, 1987); encrinurids (Best, 1961; Chatterton and Campbell, 1980; Strusz, 1980:23; trinucleids (Bertrand and Lesp^rance, 1971; Lesp€rance and Bertrand, 1976; Sheldon, 1987; Shaw, 1991); pterygometopids (Shaw, 1974:40); and phacopids (Eldredge, 1972). In the light of these results it seems likely that trilobite taxonomy is commonly too finely split and that the systematics of many taxa may be in need of substantial revision. This suggestion is not new (see Rasetti, 1948), but requires reiteration because new species continue to be errected on the basis of small numbers of specimens, with insufficient consideration of intraspecific variation. Bivariate Analysis Introduction Nominal and ordinal characters present strong evidence for the wide range of morphological variation in Dikelocephalus, but are of limited value in determining the likely controls on this variation. Quantitative comparisons of sclerite dimensions, which are discussed in this section, allow a more detailed examination of controlling mechanisms responsible for mor¬ phological variation. Bivariate analysis, by facilitating detailed examination of sclerite growth relationships, enables resolution of more subtle morphological variation. Approach. —Bivariate analyses examine the relationships between pairs of variables through growth. Labandeira’s (1983) bivariate work on Ulrich and Resser’s (1930) type suite demonstrated that each pair of variables shows a single trend line through holaspid growth. Labandeira and Hughes (1994) examined the variation of 24 variables using a combined univariate, bivariate, and multivariate approach to study Ulrich and Resser’s (1930) species, and were unable to discriminate more than one morphospecies. Labandeira (1983) did not discuss growth in detail but concluded that all the growth relationships are isometric and show little dispersion about the general trend. The work reported in this section aims to document and account for patterns of variability within Dikelocephalus. Characters used in bivariate analysis were selected from the suite of 24 characters previously studied by Labandeira and Hughes (1994) (Table 3; Figure 3). The overall pattern of morphological variability in Dikelocephalus was assessed by multivariate techniques. Methods. —Labandeira and Hughes’ (1994) study of the pooled type-suite of Dikelocephalus was not designed to reveal locality-related variation. To investigate such variation re¬ quired extensive material from well-documented sites. Al¬ though large numbers of specimens were available many proved to be unsuitable for morphometric analysis of some characters. The methods of measurement and character choice are discussed in the introduction. All reduced major axis (RMA) analyses in this section were computed using logarithms of the original measurements. A consequence of using logarithms is 24 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY over-emphasis of variation among smaller specimens. This explains the seemingly high values for the coefficients of variation in collections composed of small specimens. Standard abbreviations for sclerite dimensions were used (Table 3; Figure 3). The morphological variation in each character is assessed with reference to a standard measure for size. The standard chosen for the cranidium is the occipital-glabellar length (bl), and for the pygidium it is the intra-articulating pygidial length (zl). These dimensions were selected because axial features tend be relatively invariant in trilobites (Palmer, 1957). Results Cranidium Glabellar Width. —The overall shape of the glabella was assessed by examining the relationship between the occipital lobe width and the occipital-glabellar length. A pooled sample from five localities (including Tiinnel City Group specimens) yields a very strong correlation between these two dimensions (r = 0.993) and there is little variation in glabellar shape in the sample (Figure 15). Specimens from outside the northern Mississippi Valley fall within this range of variation. The growth of the glabella is positively allometric at the 95% significance level. There are no significant differences in the glabellar shape among collections. The growth relationships were examined using reduced major axis (RMA) and correlation coefficients. The reduced major axis of the pooled sample (log values of all specimens from all collections) shows a growth ratio of 1.076, suggesting allometric growth (p< 0.05). Comparison of the glabellar width/length ratio (k/bl) to glabellar length (bl) (Figure 16) also shows a significant positive correlation (r = 0.470, p<0.01). This suggests that slight ontogenetic increase in the relative width of the glabella may occur within the collection. This correlation could reflect the fact that collec¬ tions characterized by narrower glabellae have lower mean sizes, but there is also significant correlation between the ratio and size within individual collections. The collection with the largest size range (NF8) shows a significant correlation between the ratio and overall size (r = 0.322, p<0.05). Within Stillwater (SWb), Tiinnel City Group (TCG), and Arcadia (A Aa) collections there is an apparent tendency for the ratio to increase through growth, although these relationships are not statistically significant, perhaps due to the small sample sizes. Glabellar length/width relationships are the ones most likely to be distorted by compression and deformation, because the dimensions are orthogonal and the glabella is vaulted (Hughes and Jell, 1992). The correlation coefficient for glabellar length/width in Dikelocephalus is the highest of all pairwise correlation coefficients in the cranidium (Table 18). Hence the effects of compression appear to be slight compared to original biological variation within Dikelocephalus, provided that Table 4.— Reduced major axes for occipital lobe width/occipital-glabellar length in Dikelocephalus from four SL Lawrence Formation localities and a pooled sample from the Tiinnel City Group, (a represents the slope equation; b represents the intercept value on the y axis (in this case the y axis is the occipital lobe width [k]). NF2 = North Freedom Bed 2; NF8 = North Freedom Bed 8; A Aa = Arcadia Bed 18; SWb = Stillwater, TCG = pooled sample from Tiinnel City Group.) Locality a b All 1.076 -0.120 NF2 1.020 -0.082 NF8 1.053 -0.116 A Aa 1.068 -0.112 SWb 1.061 -0.120 TCG 1.138 -0.132 Table 5.—Coefficients of relative dispersion about reduced major axis of occipital lobe width/occipital-glabellar length. Locality Coefficient of relative dispersion NF2 4.34 NF8 8.11 A Aa 14.12 SWb 14.45 TCG 59.02 obviously deformed specimens are excluded from the data set proir to analysis. The growth allometry in glabellar shape is nearly masked by size-independent phenotypic variation within samples, as the within-collection correlation coefficients between glabellar width/length and glabellar length are low and the degree of scatter about the trend is high (Figure 16). Reduced major axes for the five collections all show similar slope values (b) (Table 4) suggesting that the growth relationship governing glabellar shape was very similar in all localities (Figures 15, 16). No collection-based differences are apparent among the St. Lawrence Formation collections (Plates 1-4, 6, 7, 9). The complete overlap at the 95% confidence level suggests that there is continuous variation throughout the sample, even though there are statistically significant differ¬ ences between the Tiinnel City Group and North Freedom Bed 2 collections. This statistical difference may be attributed to the different mean sizes of these collections. The coefficient of dispersion was calculated for each locality to assess whether localities showed different degrees of intracollection variation (Table 5). There are differences in the coefficient between the collections, but the level of intracollec¬ tion variation is relatively constant within St. Lawrence Formation localities. The greater variation within the Tiinnel City Group sample is probably due to the pooling of specimens from several localities. The morphology of the glabella is often considered conservative at low taxonomic levels in trilobites (Stubblefield, 1936; McNamara, 1986) and hence variation in glabellar form has been used extensively in the taxonomy of the Dikelo- NUMBER 79 25 localities: ■ SWb a AAa x NF2 + NF8 • TCG FIGURE 15.—Bivariate scatterplot showing relationship between occipital-glabellar length and occipital lobe width in Dikelocephalus from five localities in the northern Mississippi Valley; n = 105. Localities are Stillwater (SWb), Arcadia Bed 18 (AAa), North Freedom Bed 2 (NF2), North Freedom Bed 8 (NF8), and a pooled sample from the 'Tlinnel City Group (TCG). localities: A X + SWb AAa NF2 NF8 TCG FIGURE 16. —Bivariate scatterplot showing ratio of occipital lobe width divided by occipital-glabellar length plotted against occipital-glabellar length. Data set as in Figure 15. 26 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY cephalidae (Ulrich and Resser, 1930; Lochman, 1959). The limited variation in glabellar shape in Dikelocephalus suggests that all specimens are closely related. Ulrich and Resser (1930) made use of the relative width/length ratio in their taxonomy of Dikelocephalus (Table 1). Large species, such as D. orbicula- tus, had small glabellar length/width ratios (Ulrich and Resser, 1930:29), whereas small species, such as D. edwardsi, had larger length/width ratios (Ulrich and Resser, 1930:49). This analysis shows that these variations are not species-specific and may reflect slight allometric growth of the glabella. Labandeira (1983) assessed the relationship between occipital-glabellar length (bl) and width (k) within Ulrich and Resser’s type suite and found an almost identical correlation (r = 0.994) to that found in this study. Frontal Area Length. —Growth of the frontal area is slightly positively allometric. There is great size-independent variation in frontal area length both within and between localities but there is no justification for the recognition of more than one morphotype. Dikelocephalus from outside the northern Mississippi Valley region have frontal area lengths within the range shown by specimens in the region. The relationship between frontal area length (fl) and occipital-glabellar length (bl) was assessed using specimens from six localities within the St. Lawrence Formation and a pooled sample from the Tlinnel City Group (Table 6). The relatively low correlation coefficient for the total sample reflects the high degree of variability in frontal area length (r = 0.934) (Figures 17-19). The growth of the frontal area is positively allometric at the 95% confidence level, although the relative increase in frontal area length was all but swamped by phenotypic variation and is not particularly variable in Dikelocephalus. The reduced major axis of the pooled sample shows a specific growth ratio of 1.116, suggesting positive allometric growth (p< 0.05). However, the ratio of frontal area length to occipital-glabellar length (fl/bl) is not correlated with occipital-glabellar length (bl) (r = 0.032), which suggests that this allometry is slight. Analyses of individual collections demonstrate that only the pooled Tlinnel City Group collec¬ tions show both significant positive allometry of the frontal area (Figure 20; Plate 9: Figures 1-5, 7-11, 16). The allometry in the Tlinnel City Group collection is suprising because the specimens come from six localities. It could be an artifact of pooling the specimens or it may be a difference that exists between Dikelocephalus from the Tlinnel City Group and St. Lawrence Fonnation. Removal of the Tlinnel City Group collection from the data set does not significantly alter the correlation coefficient of the total sample. There is great variation in the length of the frontal area relative to occipital-glabellar length both within and between collections (Figures 17-19; Plate 2: Figures 1-3, 6, 7, Plate 3: Figures 1-6, Plate 4: Figures 1-9, Plate 6: Figures 1-14). Although there is a significant difference at the 95% confidence level between the collections with the greatest difference in slope equations (Tlinnel City Group and North Freedom Bed Table 6. —Reduced major axes for frontal area length/occipital-glabellar length in Dikelocephalus from six St Lawrence Formation localities and a pooled sample from the T\innel City Group, (a represents the slope equation; b represents the intercept value on the y axis (in this case the y axis is the frontal area length [fl]). NF2 = North Freedom Bed 2; NF8 = North Freedom Bed 8; A Aa = Arcadia Bed 18; SWa = Fairy Glen, Stillwater, SWb = Stillwater, LRc = Button Bluff; TCG = pooled sample from Tlinnel City Group.) Locality a b All 1.116 -0.487 NF2 0.997 -0.443 NF8 1.024 -0.403 AAa 1.020 -0.439 SWa 1.082 -0.543 SWb 1.037 -0.542 LRc 1.140 -0.617 TCG 1.284 -0.412 Table 7.—Coefficients of relative dispersion about reduced major axis of frontal area lengthy occipital-glabellar length. Locality Coefficient of relative dispersion NF2 25.17 NF8 21.68 AAa 20.20 SWa 30.66 SWb 28.84 LRc 22.99 TCG 20.26 2), there is overlap among all seven collections. This suggests that the pattern of variation of the frontal area is clinal. Variation in the ratio of frontal area length to occipital-glabellar length (fl/bl) is continuous throughout the sample and is not correlated with overall size; there are no apparent subgroups within collections. No ratio value is exclusive to a particular collection and frontal area length cannot be used to discrimi¬ nate between taxonomic groups. It is, however, possible to recognize end-members within the total range of variation and this may suggest a clinal pattern of variation in frontal area length. Reduced major axes show that some collections differ significantly in respect of frontal area length (Table 6). There are some differences in levels of intracollection variation (Table 7; Figures 17-20). Some localities show relatively constant frontal area lengths throughout the collection (i.e., Arcadia Bed 18; Figure 17), whereas others show great variation (i.e., North Freedom Bed 2 and Stillwater). Within collections, the range of the ratio of frontal area length to occipital-glabellar length (fl/bl) is similar in both small and large cranidial size classes (Figures 14-17), which suggests that variance is not size-related. In spite of the fact that there are significant differences between collections (Arcadia [AAa] and Button Bluff [LRc] collections show almost no overlap; Figure 19a), it is impossible to identify the provenance of specimens with NUMBER 79 27 CO c a> CO 0 > k_ co 2 c o localities: o SWa ■ SWb A AAa x NF2 + NF8 ♦ LRc • TCG 0.0 1.0 2.0 3.0 4.0 5.0 6.0 occipital-glabellar length: bl (cm) FIGURE 17.—Bivariate scatterplot showing relationship between occipital-glabellar length and frontal area length in Dikelocephalus from seven localities in the northern Mississippi Valley; n = 171. Localities are Fairy Glen, Stillwater (SWa), Stillwater (SWb), Arcadia Bed 18 (AAa), North Freedom Bed 2 (NF2), North Freedom Bed 8 (NF8), Button Bluff (LRc), and a pooled sample from the Tlinnel City Group (TCG). localities: □ SWa ■ SWb A AAa X NF2 + NF8 A LRc • TCG FIGURE 18.—Bivariate scatterplot showing ratio of frontal area length divided by occipital-glabellar length plotted against occipital-glabellar length. Data set as in Figure 17. confidence on the basis of frontal area length. Variation in the frontal area lengths is considerable both within and between collections. There are no grounds for recognizing particular morphs or subspecies within the sample. Dikelocephalus cranidia from Nevada, Montana, and Alberta show a range of frontal area/occipital-glabellar length ratios from 0.227 to 0.389. This range is within that shown by northern Mississippi Valley specimens (0.143 to 0.503) and there is no reason to think that the specimens from outside the region represent different taxa. The name Dikelocephalus, literally “shovel head,” indicates the prominence of the frontal area, the relative length of which has been used as an important taxonomic character. The “width of the brim” was one of the most important criteria used by Ulrich and Resser (1930:8) in the recognition of species. They established two groups of species within Dikelocephalus largely on the basis of frontal area morphology. The oweni group was characterized by broad frontal areas, and the gracilis group was characterized by short frontal areas. The analysis above shows that there is no justification for the use of this character in specific differentiation. In some Early Paleozoic trilobites the border may shorten during the holaspid phase (Hu, 1971), but the ontogenies of many primitive libristomate species show a progressive increase in the length of the frontal area relative to overall size through the meraspid and into the holaspid periods (e.g.. 28 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY occipital-glabellar length: bl (cm) localities: □ SWa ■ SWb x NF8 Figure 19.—Plots showing relationship between ratio of frontal area length dividied by occipital-glabellar length against occipital-glabellar length in Dikelocephalus from the St. Lawrence Formation. A, Specimens from Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), and Button Bluff (LRc); n = 96. B, Specimens from Fairy Glen, Stillwater (SWa), Stillwater (SWb), and North Freedom Bed 8 (NF8); n = 65. Pterocephalia concava Palmer: Palmer, 1965; Taenicephalus shumardi (Hall): Hu, 1981; Idahioa msconsinensis (Owen) and Orgymaspis (Parabolinoides) contracta (Frederickson): Wes- trop, 1986; Hundwarella personata Reed: Hughes and Jell, 1992). Dikelocephalus likewise shows slight positive allometry in the length of the frontal area. Relative lengths of the frontal area have often been considered to be of taxonomic importance in other primitive libristomate trilobites (e.g., Palmer, 1962:34; Hu, 1981) although wide variation in frontal area length is known within Late Cambrian (Longacre, 1970, fig. 8) and Early Ordovician (Fortey, 1974, fig. 3) species. The width of the frontal area in Dikelocephalus partly reflects whether the facial suture was marginal or retracted (see above). Cranidia with relatively long frontal areas are those that possess an inflated anterior border (Plate 3: Figures 2-4, Plate 6: figures 1-14). Frontal Area Width. —The width (tr.) of the frontal area (j2) shows marked variation in Dikelocephalus (e.g., Plate 2: Figures 1-7, Plate 3: Figures 1-6, Plate 4: Figures 4-8, Plate 6: Figures 1-14). The length of the glabella (bl) and frontal area width (j2) show a high positive correlation (0.980), and the variance of frontal area width both within and between most NUMBER 79 29 o ” 1.0 o> c Q> S 0.5 k. co 2 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 occipital-glabellar length: bl (cm) FIGURE 20.—Bivariate scatterplot showing relationship between occipital-glabellar length and frontal area length in a pooled sample from the TUnnel City Group (TCG); n = 11. localities is relatively low. Growth of frontal area width is positively allometric at the 95% confidence level. Patterns of locality-based variation (Figures 18, 19) in frontal area width mimic those shown by frontal area length. A correlation of the ratio of the frontal area width to occipital-glabellar length (j2/bl) with occipital-glabellar length (bl) shows a significant but small positive correlation (0.259, pcO.Ol), confirming the general pattern of positive allometry of frontal area width. This correlation may be due to the different size ranges of the specimens in collections. Those collections that have specimens showing relatively narrow frontal areas tend to have smaller mean sizes than those having wider brims (Figures 21, 22), leading to the increased proportion of narrow brimmed forms at smaller sizes. Within individual collections there is no significant correlation between size (as expressed by the occipital-glabellar length) and relative width of the frontal area. Hence, if growth of frontal area width is allometric it is only slightly so, because large sample sizes are needed to detect it. The slope values (a) are variable among localities but there are no significant differences between collections at the 95% confidence level. The range of variation within the samples is continuous and although specimens in some collections tend to have wider frontal areas than others, these differences do not provide adequate grounds for taxonomic subdivision of Dikelocephalus. Unlike the variation in frontal area length, there is overlap of the confidence limits of the reduced major axes of all collections (Table 8) and thus there is no strong suggestion of clinal variation in frontal area width. Analysis of the ratio of frontal area width/occipital-glabellar length compared to occipital-glabellar length from the seven collections shows similar patterns to those shown in frontal area lengths. This is because there is a strong positive correlation between frontal area length and width (r = 0.955), which is higher than the correlation between occipital-glabellar length and frontal area length (r = 0.934). Hence where the frontal areas are broad they also tend to be wide. However, frontal area width is most highly correlated with occipital- glabellar length (r = 0.980). This suggests that the length of the frontal area is more variable than its width and explains why although some collections show significant differences in frontal area length, they do not show significant differences in frontal area width. It appears that frontal area morphology does not vary as a strictly integrated unit. Intracollection variability was assessed using the coefficient of relative dispersion about the reduced major axis (Table 9). Tlinnel City Group (TCG) and Button Bluff (LRc) collections showed substantially higher variation than did collections from other localities. In the case of the Tlinnel City Group this may be due to the pooling of specimens from several localities, but this cannot explain the high levels of variation within Button Bluff. There are slight differences in the level of intralocality variability within the other collections. Ranking of collections TABLE 8.—Reduced major axes for frontal area width/occipital-glabellar length in Dikelocephalus from five St Lawrence Formation localities and a pooled sample from the Tlinnel City Group, (a represents the slope equation; b represents the intercept value on the y axis (in this case the y axis is the frontal width length [j2]). NF2 = North Freedom Bed 2; NF8 = North Freedom Bed 8; A Aa = Arcadia Bed 18; SWb = Stillwater, LRc = Button Bluff; TCG = pooled sample from Tlinnel City Group.) Locality a b All 1.078 0.117 NF2 1.009 0.152 NF8 1.036 0.154 A Aa 0.995 0.114 SWb 1.063 0.100 LRc 0.961 0.094 TCG 1.111 0.102 30 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY occipital-glabellar length: bl (cm) FIGURE 21.—Bivariate scatterplot showing relationship between occipital-glabellar length and frontal area width in a pooled sample of Dikelocephalus from six localities in the northern Mississippi Valley; n = 125. Localities are Stillwater (SWb), Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), North Freedom Bed 8 (NF8), Button Bluff (LRc), and a pooled sample from the Tlinnel City Group (TCG). Table 9.—Coefficients of relative dispersion about reduced major axis of frontal area width/occipital-glabellar length. Locality Coefficient of relative dispersion NF2 7.12 NF8 7.06 AAa 11.95 SWb 8.98 LRc 40.49 TCG 60.82 in terms of the levels of intracollection variability reveals substantial differences in the order obtained for frontal area length and width. This suggests that a portion of the variation in frontal area length and width is independent. Ulrich and Resser (1930) frequently used the width of the frontal area in their species designations. They viewed the frontal area as an integrated unit and characterized species on both its length and width (Ulrich and Resser, 1930:8). Raasch (1951) suggested that his revised taxa D. minnesotensis and D. oweni differed in the breadth of the frontal area. Labandeira NUMBER 79 31 occipital-glabellar length: bl (cm) FIGURE 22.—Bivariate scatterplot showing ratio of frontal area width divided by occipital-glabellar length plotted against occipital-glabellar length. Data set as in Figure 21. E o u 2.0- O) 1.5- c CD ' 1.0- a> -O • JO 0.5- 15 .o « 0.0- _Q- CO 0. T- 1 -1 - 1 - 1 -»- T 1.0 2.0 3.0 4.0 5.0 occipital-glabellar length: bl (cm) 6.0 localities: □ SWa ■ SWb A AAa X NF2 + NF8 ♦ LRc FIGURE 23.—Bivariate scatterplot showing relationship between occipital-glabellar length and palpebral lobe length in a pooled sample of Dikelocephalus from six localities in the St Lawrence Formation; n = 131. Localities are Fairy Glen, Stillwater (SWa), Stillwater (SWb), Arcadia Bed 18 (A Aa), North Freedom Bed 2 (NF2), North Freedom Bed 8 (NF8), and Button Bluff (LRc). (1983) did not discuss frontal area width. The present analysis shows that there is no justification for species designations based on frontal area morphology. Palpebral Lobe Length— The relative length of the palpebral lobe decreases during ontogeny, remaining nega¬ tively allometric throughout holaspid growth. The rate at which the eye length/occipital-glabellar length ratio decreases pro¬ gressively diminishes and the allometry is negligible at larger sizes. This trend is consistent within six collections from St. Lawrence Formation localities. Dikelocephalus from outside the northern Mississippi Valley area follow the same trend and lie within the range of variability present among northern Mississippi Valley specimens. The pooled sample from all six collections shows a curvilinear trend between palpebral lobe length (c) and occipital-glabellar length (bl) (Figure 23; Hughes, 1991, fig. 3). The ratio of the two variables changes as overall size increases. The trend line flattens through growth (Figure 23), meaning that the allometry is negative. Hence, the relative length of the palpebral lobe compared to that of the glabella 32 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY decreased through ontogeny. Reduced major axis analysis shows significant negative allometry (Table 10; also see Plate 2: Figures 1-3, 6-8, Plate 3: Figures 1-6, Plate 4: Figures 4-9, Plate 6; Figures 2-13, Plate 9: Figures 1, 3, 9, 11, 15). A logarithmic regression function describes the relationship between palpebral lobe length (c) and occipital-glabellar length (bl) (Hughes, 1991, fig. 3). This suggests that a constant growth control operated throughout the holaspid period. Correlations can only be calculated when the data are adjusted to account for the logarithmic nature of the relationship of palpebral lobe length (c) and occipital-glabellar length (bl). The correlation coefficient (r) for the pooled data set is 0.945. The ratio of palpebral lobe length to occipital-glabellar length (c/bl) is size-dependent (Figure 24). In specimens with glabellae longer than about 4 cm, the ratio stabilized to a value of about 0.3, and remained constant thereafter. Growth of the eye in cranidia with glabellae larger than 4 cm (sag.) was effectively isometric. The influence of locality on the relation¬ ship between palpebral lobe length and occipital-glabellar length was investigated using reduced major axes. Axes were calculated for the entire collection and from each individual locality to assess the significance of any locality-based variation (Table 10). The low values of the slope equation (Table 10) indicate negative allometry of the palpebral lobe, which is significant at the 95% confidence level in all collections. The two most dissimilar values are those of Arcadia (A Aa) and the Fairy Glen locality at Stillwater (SWa); these collections show a signifi¬ cant difference in growth allometry (at p<0.05). However, the differences between all pairs of most similar localities are not significant and hence the variation within the whole sample is continuous and appears clinal. There is no reason to suspect that the intercollection variation in the growth of the palpebral lobe represents more than intraspecific variation. In the bivariate plot (Figure 23) specimens from Arcadia (AAa) are concentrated along the top of the trend line (Hughes, 1991, fig. 3) although they overlap considerably with specimens from other localities. In the initial instars of holaspid ontogeny, Dikelocephalus specimens from Arcadia had relatively large palpebral lobes compared to specimens from all other localities (see Plate 3: Figures 1-6). The coefficients of relative dispersion suggest that speci¬ mens from most localities had a similar degree of variation about the general growth trend of the palpebral lobe (Table 11; Figures 23, 24). The collection from the Fairy Glen locality (SWa) shows the relatively little intracollection variation, which is surprising because this collection includes specimens showing unusually large variability in frontal area morphology. Similarly, collections from Arcadia (AAa) and North Freedom Bed 8 (NF8) show marked intraspecific variation, but show little variation in frontal area length relative to specimens in other collections. Although the palpebral lobe has been mentioned in previous studies of Dikelocephalus its allometric growth has not been noted by previous workers. The size of the palpebral lobe was often used by Ulrich and Resser (1930) to characterize their species. For example D. gracilis was considered to have “relatively much larger eyes” than D. minnesotensis (Ulrich and Resser, 1930:43). The syntype specimens used to define D. gracilis were much smaller than those assigned to D. minnesotensis. £ 0.5 H o a> 0.4 -j '35 0.89)). Principal component 2 has loadings similar to principal component 3 of the previous analysis and a similar “shift” occurs between subsequent principal components. Tunnel City Group Specimens: In order to assess the relationship of Tunnel City Group specimens to those of the St. Lawrence Formation a multiple gToup principal component NUMBER 79 45 analysis was performed, with St. Lawrence Formation and TUnnel City Group specimens defined as different groups. Only nine TUnnel City Group specimens were sufficiently well preserved to be included in the analysis, and this small sample size limits the confidence that can be placed in the results of the principal component analysis of this group. The principal components calculated did, however, accord to a large extent with patterns expected a priori and were shown to be statistically similar to those in the St. Lawrence Formation collection. As in the analysis of the St. Lawrence Formation collections, principal component 1 accounts for more than 80% of the total variation (81.9%) and principal component 2 accounts for most of the remaining variance (16.0%). The first three principal components account for 99.4% of the total variance. Patterns of character loading on the principal components were also similar to those in the St. Lawrence Formation analysis, the only notable difference being that the pattern of loading on principal components 4 and 5 were reversed. The width of the glabella (k) shows a positive loading on principal component 3, which may reflect the positive allometry of glabellar width noted in the bivariate analysis. Similarly, principal component 6 shows strong negative loading for occipital-glabellar length (bl) and positive loading for the glabellar width (k) again reflecting this allometry, although the total variance accounted for by this axis is so low that little confidence can be placed in the significance of this result. Separate principal component analyses of St. Lawrence Formation and TUnnel City Group samples were compared with a principal component analysis of the pooled sample from both units. Chi-squared tests showed that the first three principal components are not significantly different between the pooled sample and 1, St. Lawrence Formation specimens (p<0.99, p<0.98, p< 0.99 for differences between princi¬ pal components 1 to 3 respectively); and 2, TUnnel City Group specimens (p<0.64, p<0.21, p<0.13 for differences between principal components 1 to 3 respectively). This suggests that similar patterns of variation occur among both St. Lawrence Formation and Tunnel City Group specimens. Scores on the first three principal components in the pooled analysis (Figures 33, 34) show that the TUnnel City Group specimens lie within the range of variation shown by St. Lawrence Formation Dikelocephalus. Tunnel City Group specimens occupy a smaller volume of morphospace than do St. Lawrence Formation specimens, probably due to their smaller sample size. However, the specimens appear to plot on weakly defined arcs (Figures 33, 34). There may be a trend toward larger specimens showing lower scores on principal component 2 (Figure 33), but the small sample size prevents assessment of the significance of this trend. If confirmed by additional observations this pattern may reflect the unequal allometries of frontal area growth observed in the bivariate analyses of TUnnel City Group specimens (which was based on larger numbers of specimens). As frontal area length (fl) increased allometrically at a faster rate than frontal area width (j2), the angular divergence of the suture would necessarily reduce, unless there was concomitant reduction in the width of the preocular incisure, which is not observed (Plate 9: Figures 1-5,7-11). Nonmetric Multidimensional Scaling Multidimensional scaling calculates configurations of indi¬ viduals in a specificed number of multivariate dimensions, and then compares these configurations with the original associa¬ tion matrix of proximities. Where disparities exist the configuration is “stressed” to increase the goodness of fit between the configuration and the actual similiarities between individuals. This process is iterative. In this analysis three multivariate dimensions were selected and fifty iterations were computed. Stress values ranged from 14.06 at the first iteration to 1.07, an extremely low value of stress (Kruskal, 1964), at the sixth iteration, and remained constant thereafter. Considerable confidence can be placed in the resultant configuration, which was calculated on the basis of the 50th iteration because of the low stress value calculated in the last iterations. Because MDS operates on an association matrix based on similarities, all cranidia of Dikelocephalus could be included in the analysis. The configuration of individuals in the cranidial nonmetric MDS analysis accords with the results of principal component analysis (Figures 35-37). The correlation between each of the first three multivariate axes calculated in the two techniques is high, indicating a close accordance of the results of principal component analysis and non-metric multidimen¬ sional scaling (pci, MDS dimension 1 = -0.941; pc2, MDS dimension 2 = 0.944; pc3, MDS dimension 3 = -0.930). Dimension 1 of the MDS is largely size influenced, as expected. The areas of morphospace defined by each locality all overlap, when dimensions 1 and 2 are plotted (Figure 35), but TUnnel City Group specimens and those from Arcadia (AAa) and Stillwater (SWb) occupy separate area of morphospace in the plot of dimensions 2 and 3 (Figure 36). Specimens from Arcadia and those from North Freedom Bed 8 do not overlap in the plot of dimensions 2 and 3 (Figure 36), as in principal component analysis. These variations confirm the clinal pattern of variation documented by the bivariate analyses. Pygidia. —Principal component analysis. St. Lawrence Formation Specimens: A correlation matrix of the four input variables shows a significant relationship between all pairs (Table 23). The high correlation coefficients between the intra- articulating pygidial length (zl) and both pygidial width (w) and axial length (yl) imply that the overall shape of the pygidium is rather constant among specimens. The intra- articulating spine length (z5) is less strongly correlated with the other characters. The data accord well with the results of bivariate studies (Table 18), which show variation in the spine 46 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY stratigraphic unit: o SLF • TCG principal component 1 FIGURE 33.—Bivariate scatterplot of first two principal components in a sample of Dikelocephalus cranidia from the SL Lawrence Formation (SLF) and the Thnnel City Group (TCG); n = 54. All specimens considered in Figure 30 are included in the St. Lawrence Formation population. Thnnel City Group specimens are a pooled sample from five localities. eo stratigraphic unit o SLF • TCG principal component 2 FIGURE 34.—Bivariate scatterplot second and third principal components in a sample of Dikelocephalus cranidia from the northern Mississippi Valley. Data set as in Figure 33. TABLE 23.—Correlation matrix for pygidial characters of the multivariate sample of Dikelocephalus from the St. Lawrence Formation (n = 37). All values indicate statistical significance at the 99.9% confidence level. zl z5 yi w zl 0.963 0.990 0.990 z5 0.963 0.950 0.950 yi 0.990 0.950 0.991 w 0.990 0.950 0.991 Table 24.— Eigenvalues and percentage of variance accounted for by each principal component in pygidial analysis of Dikelocephalus from the SL Lawrence Formation. Principal component 1 2 3 4 Eigenvalue 3.92 0.68 0.01 0.01 Percentage variance 97.94 1.62 0.22 0.21 Cumulative percentage 97.94 99.56 99.77 100.00 NUMBER 79 47 localities: ■ SWb A AAa x NF2 + NF8 • TCG -3-2 -10 1 2 dimension 1 FIGURE 35. Bivariate scatterplot of first two multivariate dimensions calcultated by nonmetric multidimen¬ sional scaling of Dikelocephalus cranidia from the northern Mississippi Valley. Data set as in Figure 33. localities: ■ SWb A AAa X NF2 + NF8 • TCG Figure 36.—Bivariate scatterplot of second and third multivariate dimensions calculated by nonmetric multidimensional scaling of Dikelocephalus cranidia from the northern Mississippi Valley. Data set as in Figure 33. length to be a principal source of morphological variation within St. Lawrence Formation Dikelocephalus. As in the analysis of the cranidium, principal component 1 explains most of the variation (Table 24). Less than 2% is accounted for by the other principal components. The overwhelming importance of principal component 1 might swamp other styles of variation, yielding subsequent principal components that account for a tiny proportion of the total variance, with patterns of eigenweight loadings that may not be significant However, several observations that principal components 2 to 4 do represent significant patterns of variation in Dikelocephalus. Firstly, the angle between principal compo¬ nent 1 and a hypothetical vector of isometry (0.47°) is not significant (p<0.91 ), suggesting that principal component 1 reflects only differences in overall size within the sample. This is supported by the observation that the eigenweights of principal component 1 are all positive and of similar magnitude (Table 25). Intra-articulating spine length (z5) shows a strong positive loading on principal component 2, whereas other variables show negative eigenweights but of smaller magnitude. Subse¬ quent principal components reflect the influence of one or more 48 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 37.—Trivariate scatterplot of first three multivariate dimensions calculated by nonmetric multidimen¬ sional scaling of Dikelocephalus cranidia from the northern Mississippi Valley. Data set as in Figure 33. of the variables, but there is no consistent pattern of covariance after size effects have been removed. The length of the axis (y 1) has strong positive loading on principal component 3, whereas the pygidial width (w) shows negative loading, suggesting that relatively wide forms tend to have a shorter axis. Bivariate plots of the first three principal components (Figures 38, 39) show that there is complete overlap among individuals from the four localities. Several of the smaller specimens from Arcadia (AAa) show high scores on principal component 2, reflecting the extremely long spines present at this locality. Other specimens from this locality, however, show scores that fall within the range of values present among individuals from all other localities. There are no grounds for recognizing two morphs within the Arcadia population, for abundant intracollection variation in other pygidial characters from this locality is not linked with the pattern of variation in spine length. Tunnel City Groups Specimens: In order to assess the relationship of Tlinnel City Group specimens with those of the Table 25. —The relationship of each variable to each principal component is shown by the eigenweight in pygidial analysis of Dikelocephalus from the St. Lawrence Formation. Principal eigenweight (directional cosine) component zl z5 yi w 1 0.503 0.493 0.502 0.502 2 -0.137 0.856 -0.356 -0.348 3 -0.223 0.046 0.774 -0.587 4 -0.821 0.146 0.148 0.532 St. Lawrence Formation a multiple group principal component analysis was performed, with St. Lawrence Formation and Tlinnel City Group specimens defined as different groups (Figures 40, 41). Only fourteen Tlinnel City Group specimens were sufficiently well preserved to be included in the analysis, and this small sample size limits the confidence that can be placed in the results of the principal component analysis of this group (Figure 42). However, as in the analysis of the cranidia the principal components calculated did accord to a large extent with patterns expected a priori, based on the results of the bivariate analysis. Principal component 1 accounted for 98.0% of the total variation in the pooled sample. This slight increase in the amount of variation accounted for by principal component 1 compared to that of the St. Lawrence Formation specimens suggests that St. Lawrence Formation and Tunnel City Group samples show very similar patterns of variation. Principal component 1 is again isometric (p< 0.96). The first two principal components account for 99.4% of the total variance. Patterns of character loading on the principal components were also very similar to those in the St. Lawrence Formation analysis, the only notable difference being that the polarity of loading on principal component 4 was reversed in the pooled analysis. Separate principal component analyses of St. Lawrence Formation and Tlinnel City Group samples were compared with a principal component analysis of the pooled sample from both units. Chi-squared tests showed that the first three principal components are not significantly different between the pooled sample and St. Lawrence Formation specimens NUMBER 79 49 CM localities: ■ SWb A AAa X NF2 + NF8 -4-202 4 6 8 principal component 1 e -'. 1 i ■* - :/■> 1 1 ■ if % ‘‘lJ-■ ■ I ■ . I 1 H 'V- J mmr- •¥- • -I -^1 1 ■•' ^ ^ ■:*>*'. V. -V-V ."- rfvl a» - 1 IcW’jS v r'. 1 I) M ] w> : wjB L xtf jfi. : -V.’ ■ ,“r ;*«, •-/#- •*V' '?J J • *' 'tn\ ■\ r :-4x* Igfe y r „ --^ 1 ■ •$* i>*? f Vv 1 72 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 3 Dikelocephalus minnesotensis Owen: from Arcadia, Trempealeau County, Wisconsin (AAa). St. Lawrence Formation; heterolithic facies. FIGURE 1.—UW 4006-364; cranidium, latex of counterpart, with the unusually large early holaspid palpebral lobes characteristic of this locality, x2. FIGURE 2.—UW 4006-369; cranidium, latex of counterpart, with the unusually large early holaspid palpebral lobes characteristic of this locality and low anlge of divergence of the anterior branch of the facial suture, x2. FIGURE 3.—UW 4006-293; cranidium, with features typical of this locality, pustulated and with small median tubercle, xl.5. FIGURE 4.—UW 4006-372; cranidium, pustulated and with median tubercle, xl. FIGURE 5.—UW 4006-338; cranidium, xl. FIGURE 6.—U W 4006-359; cranidium, latex of counterpart, with palpebral lobes of similar dimension to similarly sized holaspid cranidia from other localities, x0.75. FIGURE 7.—UW 4006-308a; free cheek, with median suture, xl. FIGURE 8.—UW 4006-286; yoked free cheeks, lacking median suture, xl. FIGURE 9.—UW 4006-294; ventral view of hypostome, with terrace ridges, xl. FIGURE 10.—UW 4006-280; thoracic segment, from posterior part of thorax showing ornament on adaxial part of pleurae, xl. FIGURE 11.—UW 4006-282; thoracic segment, from anterior part of thorax showing ornament on adaxial part of pleurae, xl. FIGURE 12.—UW 4006-324a; pygidium, with extremely long spines characteristic of this locality and equally divided pleurae, xl.5. FIGURE 13.—UW 4006-320; pygidium, showing pleural abnormality on third pleural segment on right side of pygidium, xl.5. FIGURE 14.—UW 4006-327a; pygidium, with unequally divided pleurae, xl.5. FIGURE 15.—UW 4006-306; pygidium, with pustulated axis, xl.5. Figure 16. — UW 4006-317; pygidium, xl. NUMBER 79 1 fa 1 * gy« ' ,\v «*. i'— ** >^_ | 1 1 fc:£&r< :;';.»w I Hfr $%£$■', w&t •- 4* "^1 \ ;*■ j- jp^ - _1 74 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 4 Dikelocephalus minnesotensis Owen: from Arcadia, Trempealeau County, Wisconsin (AAa). St. Lawrence Formation; heterolithic facies. FIGURE 1.—UW 4006-305; pygidium, xl. FIGURE 2.—UW 4006-323; pygidium, counterpart, with unequally divided pleurae and relatively short posterolateral spine, xO.75. FIGURE 3.—UW 4006-310; pygidium, counterpart, with subequally divided pleurae, x0.5. Dikelocephalus minnesotensis Owen: from North Freedom Bed 2, Sauk County, Wisconsin (NF2). St. Lawrence Formation; heterolithic facies. FIGURE 4.—UW 4006-102; cranidium, latex of counterpart, with long frontal area, pustulation, median tubercle, and relatively long palpebral lobes, xl. FIGURE 5.—UW 4006-205; cranidium, with short frontal area, xl. FIGURE 6.—UW 4006-119; cranidium, latex peel of counterpart, with median tubercle, xO.75. FIGURE 7.—UW 4006-236a; cranidium, with relatively small palpebral lobes, x0.75. FIGURE 8.—UW 4006-243; cranidium, x0.75. FIGURE 9.—UW 4006-190; cranidium, xO.75. FIGURE 10.—UW 4006-390; free cheek, xl.5. FIGURE 11.—UW 4006-388; free cheek, xl. NUMBER 79 ffflf " 'I'rt 1 ’’ 1* j- '£' ^-j! ^ _/ , l “ f t-’s'.T' -3nrl fp' ,.-• /".'A:-'. r (; Tj| /'■-. J . \ fiJl '- '■• :< ' • .. •• »i- | 76 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 5 Dikelocephalus minnesntensis Owen: from North Freedom Bed 2, Sauk County, Wisconsin (NF2). St. Lawrence Formation; heterolithic facies. Figure 1.—UW 4006-385; free cheek, with median suture, x0.5. FIGURE 2.—UW 4006-389; yoked free cheek with retracted cranidial anterior border, xl. FIGURE 3.—UW 4006-437; ventral view of hypostome, with maculae, x2. FIGURE 4.—UW 4006-402a; ventral view of hypostome, lacking maculae, xl. FIGURE 5.—UW 4006-“N8"; thoracic segment, from posterior of thorax, with ornament, x2. FIGURE 6.—UW 4006-196a; partial thorax and pygidium, with equally divided pleurae, xl. FIGURE 7.—UW 4006-411; pygidium, with long posterolateral spine, xl. Figure 8.—UW 4006-168; pygidium, with subequally divided pleurae, xl. FIGURE 9.—UW 4006-180a; pygidium, xO.75. FIGURE 10.—UW 4006-178b; pygidium, xl. FIGURE 11.—UW 4006-164; pygidium, counterpart with short posterolateral spine, xl. FIGURE 12.—UW 4006-170; pygidium, counterpart with short posterolateral spines and equally divided pleurae, x0.75. FIGURE 13.—UW 4006-142; pygidium, with unequally divided pleurae, x0.5. NUMBER 79 w " i. jfik ii:’ JSjv. ’ /\ • :V ' ; / .r;-' -jfi ife&v ■ ’'Wff/v :’. tvfr '^4 ■' ; A jfii 1 Ij&iy Vy^ ifffiy' 'W^yV •? \ " If 78 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 6 Dikelocephalus minnesotensis Owen: from North Freedom Bed 8, Sauk County, Wisconsin (NF8). St. Lawrence Formation; heterolithic facies. Figure 1.—UW 4006-la; cranidium, with pustulation, x3. FIGURE 2.—UW 4006-130; cranidium, with pustulation and terracing on long frontal area, x3. FIGURE 3.—UW 4006-25a; cranidium, with pustulation, median tubercle, and deformed frontal area, x2. FIGURE 4.—UW 4006-17b; cranidium, latex peel of external mold, pustulated but lacking median tubercle, x2. FIGURE 5.—UW 4006-8; cranidium, latex peel of external mold, with pustulation and median tubercle, x2. Figure 6.—UW 4006-50a; cranidium, pustulated, x2. FIGURE 7.—UW 4006-19; cranidium, showing intercalated furrows and diminished pustulation, xl. FIGURE 8.—UW 4006-32; cranidium, pustulated and with median tubercle, xl.5. FIGURE 9.—UW 4006-13; cranidium, with pustules diminished and restricted to axial region, xl. FIGURE 10.—UW 4006-77; cranidium, xl. FIGURES 11, 12.—UW4006-52a; 11, cranidium, with median tubercle, pustulation absent, xl; 12, anterior view of cranidium, xl. FIGURES 13, 14.—UW 4006-69; 13, cranidium, xO.75; 14, anterior view of cranidium, x0.75. llfpi wmmm a m&r ' Y ‘Usd',' 1 mm ifeiM 11 \- j W-e / 80 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 7 Dikelocephalus minnesotensis Owen: from North Freedom Bed 8, Sauk County, Wisconsin (NF8). St. Lawrence Formation; heterolithic facies. FIGURE 1.—UW 4006-69; lateral view of cranidium figured previously (Plate 6: Figures 13, 14), x0.75. FIGURE 2.—UW 4006-71; cranidium, very large with short palpebral lobe (anterior of which is damaged), xO.5. Figures 3, 4.—UW 4006-386; 3, free cheek, xl; 4, lateral view of free cheek, xl. FIGURE 5.—UW 4006-394; free cheek, with median suture, xl. FIGURE 6.—UW 4006-“N17’’; free cheek, with median suture, xl. FIGURES 7, 8.—UW 4006-392; 7, free cheek, showing inosculate terraces on dorsal surface and broad doublure with straight terrace lines, x0.75; 8, lateral view of free cheek, xO.75. FIGURES 9, 10.—UW 4006-430; ventral view of hypostome, xl; 10, anterior view of hypostome, xl. FIGURES 11, 12.—UW 4006-422; ventral view of hypostome, xl; 12, lateral view of hypostome, showing wing process extending dorsally, xl. Figure 13.—UW 4006-433; thoracic segment, xl. NUMBER 79 Si ' ^ t 1 :fe| £ 11 rl j,-. ■’ V* 7- ' lii ® '• _ 1 _:___ 82 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 8 Dikelocephalus minnesotensis Owen: from North Freedom Bed 8, Sauk County, Wisconsin (NF8). St. Lawrence Formation; heterolithic facies. Figure 1.—UW 4006-106; pygidium, showing pustulated axis, x2. FIGURES 2-4.—UW 4006-87: 2, pygidium, with long posterolateral spines and subequally divided pleurae, x2; 3, posterior view of pygidium, x2; 4, lateral view of pygidium, x2. FIGURE 5.—UW 4006-85; pygidium, with pustulated axis and long posterolateral spine, x2. FIGURE 6.—UW 4006-91; pygidium, xl. FIGURE 7.—UW 4006-90a; pygidium, with equally divided pleurae and boring on doublure and dorsal surfaces, xl. FIGURES 8-10.—UW 4006-96; 8, pygidium, very large, with equally divided pleurae, x0.5; 9, lateral view of pygidium, x0.5; 10, posterior view of pygidium, x0.5. FIGURE 11.—UW 4006-97; counterpart of pygidium, with short posterolateral spine, x0.75. FIGURE 12.—UW 4006-505; counterpart of pygidium, x0.75. FIGURES 13, 14.—UW 4006-126; 13, pygidium, large with short posterolateral spine and unequally divided pleurae, x0.5; 14, detail of the terrace lines on the ventral surface of pygidium, xl. 84 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 9 Dikelocephalus of unknown specific affinity. Reno Member, Tlinnel City Group. FIGURES 1, 2.—UMPC 6660a; 1, cranidium, xl.5 (see Bell, Feniak, and Kurtz, 1952, pi. 38: fig. 4a); 2, antertior view of cranidium, xl.5. Freeburg (FG), Houston County, Minnesota. FIGURES 3-5.—UMPC 6660e, holotype of Dikelocephalus freehur^ensis, Feniak; 3, cranidium, xl (see Bell, Feniak, and Kurtz, 1952, pi. 38: fig. 4e); 4, anterior view of cranidium, xl; 5, lateral view of cranidium, xl. Freeburg (FG), Houston County, Minnesota. FIGURE 6.—UMPC 6660c; free cheek, x3 (see Bell, Feniak, and Kurtz, 1952, pi. 38: fig. 4c). Hell Hollow (HW), Houston County, Minnesota. FIGURE 7.—UMPC 9402e; cranidium, x3. Hell Hollow (HW), Houston County, Minnesota. FIGURE 8.—UMPC 9402c; cranidium, x3. Hell Hollow (HW), Houston County, Minnesota. FIGURES 9-10.—UMPC 9402b; 9, cranidium, x2; 10, anterior view of cranidium, x2. Hell Hollow (HW), Houston County, Minnesota. Figure 11.—-UW 4006-256; cranidium, xl. Newton (NNa), Vernon County, Wisconsin. FIGURE 12.—UW 4006-258; ventral view of hypostome, x3, Newton (NNa), Vernon County, Wisconsin. FIGURE 13.—USNM 473967; hypostome, x2. Bean Hollow (BN), Vernon County, Wisconsin. Figure 14.—UW 4006-272; cranidium, xl. Lansing (LSb), Allamakee County, Iowa. Figure 15.—UW 4006-273; cranidium, xl. Lansing (LSb), Allamakee County, Iowa. Figure 16.—UW 4006-270; cranidium, x2. Exact locality unknown, perhaps Newton (NNa), Vernon County, Wisconsin. FIGURES 17-19.—UMPC 6660f; 17, latex peel of external mold of pygidium, xl.5 (see Bell, Feniak, and Kurtz, 1952, pi. 35: fig. 4); 18, posterior view of latex peel of pygidium, xl.5; 19, lateral view of latex peel of pygidium, xl.5. Freeburg (FG), Houston County, Minnesota. FIGURE 20.—UMPC 6600b; pygidium, x2 (see Bell, Feniak, and Kurtz, 1952, pi. 38: fig. 4b). Freeburg (FG), Houston County, Minnesota. E£& NUMBER 79 $3* m i 1 gw $ib lifl fiidM warn H V Wjv ’ : w Us L •' 4 i« |p fMmm «BHByfc #|§||I§J \ Mm e* / S \- .\»l# 86 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 10 Dikelocephalus of unknown specific affinity. Reno Member, T\innel City Group. FIGURE 1.—UW 4006-252; pygidium, xl.5. Newton (NNa), Vernon County, Wisconsin. FIGURE 2.—UW 4006-271; latex peel of pygidium, xl.5. Newton (NNa), Vernon County, Wisconsin. FIGURE 3.—UW 4006-256; latex peel of pygidium, xl.5. Newton (NNa), Vernon County, Wisconsin. FIGURE 4.—UW 4006-255; latex peel of pygidium, xl.5. Newton (NNa), Vernon County, Wisconsin. Figure 5.—UW 4006-257; latex peel of pygidium, xl.5. Newton (NNa), Vernon County, Wisconsin. FIGURE 6.—UW 4006-253; pygidium, xl. Newton (NNa), Vernon County, Wisconsin. FIGURE 7.—UW 4006-506; pygidium, x3. Newton (NNa), Vernon County, Wisconsin. FIGURE 8.—UMPC 9405a; pygidium, x3. Hell Hollow (HW), Houston County, Minnesota. FIGURES 9-11.—UW 4006-275; 9, pygidium, xl; 10, posterior view of pygidium, xl; 11, lateral view of pygidium, xl. Lansing (LSb), Allamakee County, Iowa FIGURE 12.—UW 4006-274; pygidium, xl. Lansing (LSb), Allamakee County, Iowa FIGURE 13.—UW 4006-463; pygidium, x0.75. Excelsior (EX), Richland County, Wisconsin. Dikelocephalus minnesotensis Owen: cranidia showing caeca from LaGrange Mountain (RWa), Goodhue County, Minnesota FIGURE 14.—AMNH 44020; cranidium, x2. FIGURE 15.—USNM 474688; cranidium, x2. tmmmn ■$?«?Wt HI - : ; : *;i- •■'? ai i$Xk>*0 r V« I'&iP'&lK ■iWWmi IMUp fttiy^ 88 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY PLATE 11 Dikelocephalus minnesotensis Owen: from various localities in the St. Lawrence Formation; heterolithic facies. FIGURE 1.—USNM 474689; anterior of cranidium, xl .5. 4 km north of Stoddard (near Victory), Vernon County, Wisconsin. Figures 2-5. —UW 4004-lb; 2, cranidium, xl; 3, two cranidia and counterpart of large pygidium, x0.5; 4, marginal rim on pygidium, xl; 5, post-axial emargination on pygidium, xl. Galesville (GE), Trempealeau County, Wisconsin. Dikelocephalus minnesotensis Owen: from St. Lawrence Formation; laminated sandstone subfacies. FIGURE 6.—USNM 72711; cranidium, x2. Myers Hill (MY), Monroe County, Wisconsin. FIGURE 7.—USNM 58623; cranidium, xl.5. Osceola (OA), Polk County, Wisconsin. Dikelocephalus minnesotensis Owen: from St. Lawrence Formation; heterolithic facies. FIGURES 8, 9.—AMNH 39094; 8, pygidium, x0.5; 9, posterior view of pygidium, x0.5. Madison (MS), Wisconsin. Dikelocephalus minnesotensis Owen: from outside of the northern Mississippi Valley. FIGURE 10.—USNM 447018; pygidium, xl. Shale next to a normal fault of 1 m throw, 15 m east of curve sign on hillside at highway, about 100 m east of footbridge, 4.5 km east of footbridge at Glenwood, Glenwood Springs area, Colorado. FIGURE 11.—USNM 447017; cranidium, xl. Highland Peak Quadrangle, Pioche District, Nevada. Figure 12.—GSC 75188; pygidium showing long posterolateral spines. x2 (see Westrop, 1986, pi. 3: fig. 7). Mistaya Formation, southern Alberta Site 260.2. FIGURE 13.—GSC 75208; pygidium, showing unequally divided pleurae, xl. Mistaya Formation, southern Alberta Site 260.2. FIGURE 14.—USNM 447016; cranidium showing pustulation, x3. From “0.5 km above mouth of junction of ravine (entering) from east in river beds, Hudson Creek, Tfexas.” Figure 15.-—USNM 447015; cranidium, x3. From “0.5 km above mouth of junction of ravine (entering) from east in river beds, Hudson Creek, Tfexas.” FIGURE 16.—USNM 482957; pygidium, x3. From northeast slope of Mount Lincoln, in limestone float above fault. Snake Range, Nevada. Type specimens of Dikelocephalus minnesotensis Owen: from Stillwater (SWa), Washington County, Minnesota. FIGURE 17.—USNM 447020; paralectotype cranidium, xl (see Owen, 1852, pi. 1: fig. 1). FIGURE 18.—USNM 17863; lectotype pygidium, with abnormal division of the right posterior pleurae, x0.5 (see Owen, 1852, pi. 1: fig. 1). NUMBER 79 vKlV y>.' 1 S3^ ■ ^ -^Vv ,: Bp* < ;iw $f IsfflH / I /l • • h c • w , r i \ • ^ 1 I ■ N. M j _•>_ £>*• - r , n » *» /.-••, 1 r%#p * *&;.?. r lH^f '‘,<£PV ■’ >: ■ ji Jr [ »s ..: jU4 B™ "™l( HH ^V.’^r -A-J .-> REQUIREMENTS FOR SMITHSONIAN SERIES PUBLICATION Manuscripts intended for series publication receive substantive review (conducted by their originating Smithsonian museums or offices) and are submitted to the Smithsonian Institution Press with Form SI-36, which must show the approval of the appropriate authority designated by the sponsoring organizational unit. Requests for special treatment—use of color, foldouts, case-bound covers, etc.—require, on the same form, the added approval of the sponsoring authority. Review of manuscripts and art by the Press for requirements of series format and style, completeness and clarity of copy, and arrangement of all material, as outlined below, will govern, within the judgment of the Press, acceptance or rejection of manuscripts and art. Copy must be prepared on typewriter or word processor, double-spaced, on one side of standard white bond paper (not erasable), with VW margins, submitted as ribbon copy (not carbon or xerox), in loose sheets (not stapled or bound), and accompanied by original art. Minimum acceptable length is 30 pages. Front matter (preceding the text) should include: title page with only title and author and no other information; abstract page with author, title, series, etc., following the established format; table of contents with indents reflecting the hierarchy of heads in the paper; also, foreword and/or preface, if appropriate. First page of text should carry the title and author at the top of the page; second page should have only the’ author’s name and professional mailing address, to be used as an unnumbered footnote on the first page of printed text. Center heads of whatever level should be typed with initial caps of major words, with extra space above and below the head, but no other preparation (such as all caps or underline, except for the underline necessary for generic and specific epithets). Run-in paragraph heads should use period/dashes or colons as necessary. Tabulations within text (lists of data, often in parallel columns) can be typed on the text page where they occur, but they should not contain rules or numbered table captions. Formal tables (numbered, with captions, boxheads, stubs, rules) should be submitted as carefully typed, double-spaced copy separate from the text; they will be typeset unless otherwise requested. If camera-copy use is anticipated, do not draw rules on manuscript copy. Taxonomic keys in natural history papers should use the aligned-couplet form for zoology and may use the multi-level indent form for botany. If cross referencing is required between key and text, do not include page references within the key, but number the keyed-out taxa, using the same numbers with their corresponding heads in the text. Synonymy in zoology must use the short form (taxon, author, year:page), with full reference at the end of the paper under "Literature Cited.” For botany, the long form (taxon, author, abbreviated journal or book title, volume, page, year, with no reference in “Literature Cited") is optional. Text-reference system (author, year:page used within the text, With full citation in “Literature Cited” at the end of the text) must be used in place of bibliographic footnotes in all Contributions Series and is strongly recommended in the Studies Series: “(Jones, 1910:122)" or “...Jones (1910:122).” If bibliographic footnotes are required, use the short form (author, brief title, page) with the full citation in the bibliography. Footnotes, when few in number, whether annotative or biblio¬ graphic, should be typed on separate sheets and inserted immedi¬ ately after the text pages on which the references occur. Extensive notes must be gathered together and placed at the end of the text in a notes section. Bibliography, depending upon use, is termed “Literature Cited," “References,” or “Bibliography.” Spell out titles of books, articles, journals, and monographic series. For book and article titles use sentence-style capitalization according to the rules of the language employed (exception: capitalize all major words in English). For journal and series titles, capitalize the initial word and all subsequent words except articles, conjunctions, and prepositions. Transliterate languages that use a non-Roman alphabet according to the Library of Congress system. Underline (for italics) titles of journals and series and titles of books that are not part of a series. Use the parentheses/colon system for volume (number):pagination: “10(2):5-9." For alignment and arrangement of elements, follow the format of recent publications in the series for which the manuscript is intended. Guidelines for preparing bibliography maybe secured from Series Section, SI Press. Legends for illustrations must be submitted at the end of the manuscript, with as many legends typed, double-spaced, to a page as convenient. Illustrations must be submitted as original art (not copies) accompanying, but separate from, the manuscript. Guidelines for preparing art may be secured from the Series Section, SI Press. All types of illustrations (photographs, line drawings, maps, etc.) may be intermixed throughout the printed text. They should be termed Figures and should be numbered consecutively as they will appear in the monograph. If several illustrations are treated as components of a single composite figure, they should be designated by lowercase italic letters on the illustration; also, in the legend and in text references the italic letters (underlined in copy) should be used: “Figure 9b.” Illustrations that are intended to follow the printed text may be termed Plates, and any components should be similarly lettered and referenced: "Plate 9b." Keys to any symbols within an illustation should appear on the art rather than in the legend. Some points of style: Do not use periods after such abbrevia¬ tions as “mm, ft, USNM, NNE.” Spell out numbers “one” through “nine" in expository text, but use digits in all other cases if possible. Use of the metric system of measurement is preferable; where use of the English system is unavoidable, supply metric equivalents in parentheses. Use the decimal system for precise measurements and relationships, common fractions for approximations. Use day/month/ year sequence for dates: “9 April 1976.” For months in tabular listings or data sections, use three-letter abbreviations with no periods: “Jan, Mar, Jun,” etc. Omit space between initials of a personal name: “J.B. Jones.” Arrange and paginate sequentially every sheet of manuscript in the following order: (1) title page, (2) abstract, (3) contents, (4) foreword and/or preface, (5) text, (6) appendices, (7) notes section, (8) glossary, (9) bibliography, (10) legends, (11) tables. Index copy may be submitted at page proof stage, but plans for an index should be indicated when the manuscript is submitted. Jpflwk .JzjjBHP ’ wty « \ x ,\ 1 wv Swt * hi£ VjgmTvV/ 9| , t^. \\ \y.| '- ;'• j tBi-. vV vSL W \\,i) vim »11I i 1 j. It nS WSjy* yiUk f fiffi -i -;- -