Stephen Jay Gould Source: Journal of Paleontology, Vol

Home , Succinea

Precise but Fortuitous Convergence in Pleistocene Land Snails from Bermuda Author(s): Stephen Jay Gould Source: Journal of Paleontology, Vol. 45, No. 3 (May, 1971), pp. 409-418 Published by: SEPM Society for Sedimentary Geology Stable URL: http://www.jstor.org/stable/1302686 Accessed: 01-04-2015 22:46 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

SEPM Society for Sedimentary Geology is collaborating with JSTOR to digitize, preserve and extend access to Journal of Paleontology. http://www.jstor.org

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions JOURNAL OF PALEONTOLOGY

A publication of THE SOCIETYOF ECONOMICPALEONTOLOGISTS AND MINERALOGISTS and THE PALEONTOLOGICALSOCIETY

VOLUME 45 MAY 1971 NUMBER 3

JOURNALOF PALEONTOLOGY,v. 45, NO. 3, P. 409-418, 4 TEXT-FIGS., MAY 1971

PRECISE BUT FORTUITOUS CONVERGENCE IN PLEISTOCENE LAND SNAILS FROM BERMUDA STEPHEN JAY GOULD Harvard University, Cambridge,Massachusetts

ABSTRACT-Twoconvergences between subgenera of Bermudian land snails [P. (Poecilo- zonites) and P. (Gastrelasmus)] were initiated when a taxon of one subgenus evolved to the very different size range of the other. Entrance into this new size range produced or required a set of correlated changes in shape, color, and thickness that resulted in very precise convergences. Some of these ancillary changes were direct adaptationsto the new size range: the large taxon of the generally small subgenus evolved a triangular dome; the rapid increase of height/width that produced a strong dome at the usual small sizes could not be maintainedto this large size. Other changes were fortuitous consequencesof growth and its interaction with the developmentalrate: the small taxon of the generally large subgenus retained its color bands; the small subgenus does not usually produce bands-adult shells display the blotched pattern of an earlier ontogenetic stage common to the whole genus; but one taxon of the small subgenus grew more than the usual number of whorls, forming thereby both the high dome and the color bands that produced convergence. Convergence would seem awesome if we recognized: a) how many characters can be modified together as mechanicalor developmentalconsequences of a primary adaptation; b) how limited are the adaptive solutions to common problems-in particular to altered size.

INTRODUCTION made of these terms unless "fortuitous" conver- in- BASHFORD DEAN, Columbia's eminent icthyol- gence covers all similarities that are evolved different but are not ogist, once wrote a "chapter in un-natural dependently by lineages, the results of selection for history" in which he catalogued a set of strictly operating directly their attainment. I will here that two fortuitous resemblances between, for examples, argue very in Bermudian land snails Japanese crab carapaces and oriental faces, or precise convergences include fortuitous feattires evolved as the cor- whale "earbones" beached in Norway and Scan- relates of a encroach- dinavian sailors (Dean, 1908). These examples primary adaptation-the are at one clear extreme of the continuum be- ment by an unusual form upon the size range of a related As Dean's tween "fortuitous" and "adaptive" convergences subgenus. examples show, our sense of wonder is in direct that Rudwick (1965) recognizes. Despite some inspired propor- tion to the extent of fortuitous resemblance. disagreement,' I do not see what sense can be A convergence may be "explained" at various 1Although noble attempts have been made to re- levels of satisfaction. Ideally, we know the sig- solve, or at least to categorize, the confusion (Haas nificance of each feature and the cause and Simpson, 1946), this little corner of evolutionary adaptive theory is wallowing in a mire of terms used in dif- of every fortuitous resemblance. Often, here in- ferent ways by various authors. George (1962, p. 22), for example, terms "accidental"all convergence that does not have genetic equivalence as its basis; prac- separate lineages by the selection of homologous tically every case of our adaptive-and therefore not genes). An event, of course, is "fortuitous" only in fortuitous-convergence would be termed accidental respect to the theory under which it is analyzed. The by George. (Most authors would resolve this dilemma similarity of shells and faces-or of staurolitecrystals by definition-by designating as parallelism, rather and the true cross-was once awarded great signifi- than as convergence, all similar features evolved in cance as a sign of divine unity in the plan of creation. 409

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions 410 STEPHEN JAY GOULD

eluded, we can only identify the adaptation, but genus and only one recently studied (Gould, can explain other resemblances as necessary 1969), is generally large (20-40 mm in width), correlates to it. Raup (1968) achieved this level strongly domed, thick, and colored with chest- of explanation when he rendered the form of nut-brown bands as an adult. (Band number, some echinoid plates as a consequence of close width and position vary among taxa-Gould, packing (mechanical pressure), given an initial 1969, pp. 429-431). Subgeneric terminology plate shape, rate of growth and rate of supply, was erected by Pilsbry (1924) who established but did not explain why one set of "givens" was the subfamily Poecilozonitinae (in the Zoniti- selected over another. dae) for this genus alone. b) P. (Gastrelasmus) is small (5- ARGUMENT: TWO CONVERGENCES generally 10 mm in width), strongly domed, um- AND THEIR CAUSES narrowly bilicate, thin, and colored with discontinuous I shall state the argument and then develop blotches that become stronger and wider with its points. growth but do not coalesce to form bands. It has 1. Shells of Poecilozonites (Poecilozonites) one unique and defining feature: an internal pa- are generally large (up to 45 mm in width), latal lamina that extends from the aperture to while those of Poecilozonites (Gastrelasmus) the termination of the protoconch (though it is are usually small (rarely more than 10 mm). often resorbed in the apical whorls). Were it Once during the Pleistocene history of each not for this lamina, I would not have recognized subgenus, a single form evolved to the size the convergences reported here. range of the other. c) P. (Discozonites) is generally small (8- 2. When it entered this new size range, each 15 mm in width), flat to strongly domed, thin, form converged in shape and color upon a taxon colored with discontinuous blotches, and as de- of the other subgenus. fining features, has a rounded whorl periphery 3. For the large taxon of P. (Gastrelasmus), and wide umbilicus. Both convergences are be- convergence in shape was a necessary conse- tween P. (Poecilozonites) and P. (Gastrelas- quence of growth to an unusually large size. mus); P. (Discozonites) shall not be considered The small taxon of P. (Poecilozonites) evolved further, although it developed some equally out- a shape that it could not have maintained when standing convergences with P. (Gastrelasmus). large and, thereby, converged upon a subspecies This paper reports only those convergences that of P. (Gastrelasmus). Size and shape are cor- result from the transgression of one subgenus related in each case. upon the characteristic size of another. 4. Each taxon retained, at its new converging 2. Two between and unusual size, the characteristic color pat- convergences P.(Poecilozonites) and (P.Gastrelasmus) tern of its subgenus. The taxa converged upon had evolved these same patterns. Since color Case 1 (Text-fig. 1): In all Bermudian for- has the same ontogeny in all Poecilozonites, dif- mations, the most common species of P. (Gas- ferences among taxa result from variation in trelasmus) is P. (G.) circumfirmatus Redfield; size, whorl number and rate of growth. Conver- it is depicted on the lower right of Text-figure gences in color are the fortuitous results of similarities in these variables. IL 1. The genus Poecilozonites in Bermuda

4'I The speciation of Poecilozonites on Bermuda provides a classic example of "island evolution." With the exception of Succinea, it is the only large pulmonate that reached this isolated island. It diversified remarkably and had split into three subgenera (with at least 10 species) when fossils are first found in the Pleistocene (see Pilsbry, 1924 and Gould, 1969 on Poecilozonites; Bretz, 1960 and Land, Macken- zie and Gould, 1967 on Bermudian geology; the volcanic base of Bermuda has been TEXT-FIG.1-The convergence of P.(P.) bermudensis recently and acutissinus as in Each sub- paedomorphs (upper left) P.(G.) dated Oligocene-Miocene age). (upper right). Convergence involves size, shape, genus has its characteristic size, shape, and color, and thickness. The internal palatal lamina of color. P. (Gastrelasmus) distinguishes the subgenera. Ac- a) P. (Poecilozonites), the nominate sub- tual size.

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions CONVERGENCE IN BERMUDIAN LAND SNAILS 411

1. Other taxa are larger, more widely umbilicate, or more triangular in cross-section but these variations are minor compared with that found in P. (G.) acutissimus Pilsbry, 1924. Pils- bry's types and paratypes do not stand out among the subgenus; for while they are more delicate, sharply keeled and triangular than any others, they are not distinct in size, the largest \ 11 mm in width. In 1966, at an / being exposure of the Harrington-Pembroke Formations (east of Pink Beach on the south coast), I found a sample that shares all the characters of Pils- P. acutissimus but contains bry's (G.) speci- TEXT-FIG.2-The convergence of P.(P.) cupula dalli mens that are almost twice as large (up to 20.5 (upper left) and P.(G.) circumfirmatuscavespiram mm in width). Zonitoids grow throughout life; (upper right). Convergence involves size, sh: pe, they have no recognizable adult stage. All but 3 color, and thickness. The internal palatal lamina of of are the others P. (Gastrelasmus) distinguishes the subgenera. Pilsbry's specimens broken; Actual size. may be juveniles. The extremely delicate shell is rarely preserved intact. Thus, Pilsbry missed the outstanding characteristic of P. (G.) acutis- any other P. (Poecilozonites) and had a re- simus-its large size. At widths greater than 20 markably high and strongly-domed spire. In the mm, it lies well in the range of P. (Poecilozon- 1920's, Sayles collected a large sample from ites) where I would have placed it without hesi- Harrington-Pembroke outcrops on Bermuda's tation had it not been for its palatal lamina. A south shore. I collected more specimens and specimen is shown in the upper right of Text- demonstrated that this taxon is a subspecies of figure 1. polytypic P. (P.) cupula (in Gould, 1966a, Even at its large size, P. (G.) acutissimus 1969). does not approach the characteristic form of P. P. (P.) cupula dalli is shown in the upper left (Poecilozonites). Its rather triangular cross- of Text-figure 2; its progenitor P. (P.) cupula section, wide umbilicus, thin shell and blotched cupuloides occupies the lower left. The size dif- coloration, contrast with the strong dome, nar- ference is not as striking in two-dimensional row umbilicus, thicker shell and color bands of representation as it is in actual specimens. This most P. (Poecilozonites)-lower left of Text- is a common psychological artifact. We intui- figure 1. Yet, four times during its Pleistocene tively judge size by volume when holding actual history, P. (P.) bermudensis zonatus evolved specimens. Size differences seem much smaller paedomorphic subspecies that are scaled-up rep- in figures-where we compare areas at best, and licas of its juvenile shell (Gould, 1968, 1969, pp. often only lengths. There is a large literature on 469-483). And these paedomorphic shells are: the relationship between perceived and actual a) rather triangular in cross-section-since differences in sizes, sounds, light intensities, etc. the dome develops gradually during growth. (Stevens and Girao, 1963; Stevens, 1968, for b) widely umbilicate-since umbilical width example). The principle is well understood by is a mechanical correlate of the shell's relative map makers in their choice of symbols for rela- width (Gould, 1968). tive sizes of cities or production of commodi- c) thin-since the shell increases in relative ties; but it is rarely recognized by biologists or thickness during growth. geologists. Likewise, in Text-fig. 1, P. (G.) acu- d) blotched in color-since the development tissimus is not quite twice as long as P. (G.) of color progresses from juvenile blotches to circumfirmatus: yet its volume is almost eight adult bands. In other words, these paedomorphic times as great. subspecies of P. (P.) bermudensis share all ma- Although P. (G.) cupula dalli lies in the size jor features of size, shape, color, and thickness range of P. (Gastrelasmus), it is not close to with P. (G.) acutissimus, but lack a palatal lam- the usual shape of that subgenus, for its spire is ina (upper left of Text-fig. 1). too high, its dome too strong, and it has retained Case 2 (Text-fig. 2): In 1904, Gulick de- the adult color bands of larger P. (Poecilozon- scribed, as P. dalli, a peculiar shell from an un- ites). But in 1965, I collected from the Shore known locality on Bermuda. It clearly belonged Hills soil of St. George's Island (Locality 48 of to P. (Poecilozonites) and stood in closest rela- Gould, 1969), a sample that I labelled as P. (P.) tion to P. cupula, the smallest species of this cupula dalli until I noticed the well-developed subgenus. Yet, it was a good deal smaller than palatal lamina of P. (Gastrelasmus). The char-

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions 412 STEPHEN JAY GOULD

imposed. The embryonic shell, or protoconch, is deposited within the egg, essentially as a single unit. Although it differs in mode of growth from later shell and is subject to other selective b and mechanical pressures, it nevertheless forms a nucleus about which the accretionary shell grows. Its initial shape has a great effect upon final form when growth of the dome follows a power function. Shells that differ only in initial shape and not in intensity of doming will plot as a series of parallel lines on log-log paper Text- fig. 3). On arithmetic coordinates, the same for- mulae yield curves that differ greatly in spire height at any given size. (Intensity or rate of 1 3 10 10 15 doming is not equivalent to spire height, though two are These arithme- TEXT-FIG. 3--Left. Log-log plots of spire height the usually correlated.) (ordinate) vs. spire width (absciss;a) for two tic curves are actual portrayals of spire geome- hypothetical individualsof the same dcoming inten- try, as if the shell were placed upside down with sity but different initial shapes. Lower equation: its apex near the origin of the graph. [Here, for = .05x2. of line: = y Equation upper y reasons outlined in Gould, p. Right. The same two equationson a Orith 1969, 426, equation ordinates to show that intensity of doming and (1) was fit for each shell from whorl 2 to the spire height are not synonymous. Li:ne a (lower end of growth and protoconch height was sub- line of log-log plot) has the same domling intensity tracted from each of the height measures. Thus, but lower initial shape than line b. Tl ie arithmeti measure of initial as in Ta- curves represent actual spire geometry (of an in- my shape, reported verted shell with its apex at the orrigin of the ble 1, is spire height at whorl 2-protoconch graph). Thus, the shell representedb y curve a is spire/spire width at whorl 2]. relatively wider than that of curve b at any adult c) the size at which doming begins. Since the size. height/width ratio of the spire increases contin- ually during accretionary growth, a shell that acteristic Shore Hills P. (Gastre ,lasmus) is begins doming at the same intensity and initial widely-umbilicate, thin-shelled P. (( 7.) circum- shape but at a smaller size than another will firmatus caliban. The shells at Locaility 48 are have a higher spire at any later common size. not only thick and narrowly-umbilica te (in addi- Since width at the end of the first whorl is a tion to the spire and color characterrs that con- redundant measure of protoconch width in Poe- verge on P. (P.) cupula dalli), but tiiey also oc- cilozonites (Gould, 1969, p. 443), protoconch cupy a geographic area inhabited by no other P. width is an appropriate measure of initial size; (Gastrelasmus) at that time. Theyr therefore it is recorded in Table 1. merit recognition as a subspecies, wl iich I name When intensity of doming, initial shape and P. (G.) circumfirmatus cavespiram ( Text-fig. 2, initial size are specified, height/width ratios and upper right), in reference to the dec:eptive con- spire shape can be calculated for any later size. figuration of the spire (formal desc ription and Their values, in fact, place constraints upon figure in appendix). final size. The most important of these is imposed by intensity of doming, for the higher the 3. The correlation size and. of shape k-value in (1), the faster height increases than The basic shape of a domed shell can be ren- width. High values of k cannot be maintained dered by three factors: over a wide size range, lest the spire reach ina- a) the intensity of doming. To produce a daptive heights. The negative correlation of dome, the shell spire must increas(e in height final size and intensity of doming in P. (Poeci- faster than it grows in width. If y a nd x repre- lozonites) (Gould, 1966a) reflects this princi- sent, respectively, the height and w idth of the ple: P. (P.) cupula must remain small, for ex- spire at each postprotoconch whorl, the rate of trapolation of its high k-values to the shell doming can often be represented by a standard width of P. (P.) nelsoni would produce spires power function of the form up to twenty times higher than wide. The of our = shapes convergent forms are also y bxk (1) related to their unusual sizes: in which k is the ratio of specific g rowth rates of y vs. x (Gould, 1966a, 1969, pp. 4t26-428). Case 1: For the small sizes that characterize b) the initial shape upon which doming is all but one of its species, high k-values pose no

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions CONVERGENCE IN BERMUDIAN LAND SNAILS 413

TABLE --Summary of doming factors in selected taxa of Poecilozonities

Initial size Number of Taxon k-value Initial shape (protoconch samples (20 shells per width) sample) P. (P.) bermudensispaedomorphs P. (P.) b. fasolti 2.02 .160 2.00 1 P. (P.) b. siegmundi 1.99 .250 2.13 1 P. (P.) b. sieglindae 1.99 .209 2.18 2 P. (P.) b. bermudensis 1.92 .269 2.09 10 P. (G.) acutissimus Pink Beach Sample 1.92 .173 1.65 1 Type Sample 1.77 .241 1.51 1 (7 specimens) P. (P.) cupula dalli 2.95 .418 1.52 1 P. (P.) cupula cupuloides 2.15 .266 1.85 6

problems for P. (Gastrelasmus). Most of its the range of P. (Gastrelasmus); I cannot ar- taxa have k-values greater than 2.3, relatively gue, as I did for Case 1, that the unusual shape high initial shapes, and small initial sizes; this of this convergent form was a prerequisite to its combination is strongly size-limiting. To reach entrance into a new size range. The correlation the size of P. (G.) acutissimus, some doming of size and shape is weaker here. The small size factor had to be modified. Table 1 shows that of P. (P.) cupula dalli allowed it to attain a the initial size of P. (G.) acutissimus remained shape that no normal taxon of its subgenus small; hence it needed a more triangular cross- could maintain at the usual large size. section (= lower rate of doming) in order to P. (G.) circumfirmatus cavespiram produced reach its final size. The two samples of P. (G.) its high spire both by increasing its k-value and, acutissimus display different strategies for at- more importantly, by maintaining that intensity taining large sizes. The type sample has main- of doming over a greater than usual number of tained a high initial shape, but reduced its k- whorls. value below that recorded for any P. (Poecilo- In both cases, I suspect that new sizes and zonites); the Pink Beach sample has a much shapes were produced by active selection for wider initial shape and can maintain a greater them. Beyond this, we can only speculate on pri- rate of doming. The four paedomorphs of P. mary effects and secondary consequences. In P. (P.) bermudensis have combinations of k-value (G.) acutissimnus, the low, triangular dome and initial shape that should produce spires might have evolved before size increased. Alter- higher than those of P. (G.) acutissimus, but nately, selection pressure for this shape may their initial sizes exceed those of P. (G.) acu- have arisen only when the size that required it tissimus and near-identity of shape is main- also became advantageous. At least, increase in tained by any common adult size because this size could not have preceded the modification of tendency towards higher spires began at a shape. We encounter the reverse situation in P. larger size. With its retention of small initial (P.) cupula dalli; here alteration of shape could size, convergence by P. (G.) acutissimus upon not have preceded the diminution of size. In this the shape of P. (P.) bermudensis paedomorphs case, either size and shape were selected in con- became an inevitable consequence of evolution cert, or small size came first and permitted a towards the larger sizes of P. (Poecilozonites). positive selection for high k-values with intense Case 2: As the smallest taxon of its subgenus, domes at small initial sizes. I have no way of P. (P.) cupula dalli stands at one extreme of a distinguishing these alternatives. pervasive correlation between size and shape. Its But there is also a fortuitous aspect to these mean k-value of 2.95 is the highest I have re- convergences in shape. With its small initial corded for the genus; since this is imposed on a size, P. (G.) acutissimus could only have grown high initial shape and small initial size, all dom- large by evolving a weakly-domed, rather trian- ing factors combine to produce the highest and gular spire. Yet, this shape produces no conver- most intensely-domed spire in Poecilozonites. In gence with any normal taxon of P. (Poecilozon- evolving to the small size of P. (Gastrelasmus), ites); these are all more intensely domed. The P. (P.) cupula dalli did not cease growth at unusual P. (P.) bermudensis paedomorphs have fewer whorls, but rather miniaturized its a weakly-domed, triangular spire because selec- whorls, attaining a protoconch size well within tion for a retarded developmental rate pro-

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions 414 STEPHEN JAY GOULD

longed juvenile shapes to adult sizes. Yet this sented by precursor blotches at the sixth (pae- retardation was selected to produce a thin shell domorphic P. (P.) bermudensis subspecies). in the calcium-poor environments of Bermuda's The processes of growth and development are red soils (Gould, 1968, 1969). The weak color partly correlated and partly independent. The and triangular spires are secondary, and non- ontogenetic color sequence, as an aspect of de- adaptive (though obviously not inadaptive), velopment, is obviously displayed through consequences of this primary advantage. Thus, growth-its stages, in any single individual, ap- the weakly-domed triangular spire is directly pear sequentially as size increases. But develop- adaptive in one taxon, but produced as a devel- ment can be retarded or accelerated relative to opmental consequence of another adaptation in growth, producing the phylogenetic phenomena the second. The similarity is fortuitous. of heterochrony. The paedomorphic subspecies Likewise, in becoming small and high-spired, of P. (P.) bermudensis are greatly retarded in P. (P.) cupula dalli did not resemble any nor- development; the colors and shapes that charac- mal taxon of P. (Gastrelasmius), for these are terize small ancestral juveniles do not appear all lower-spired. If the unusual P. (G.) circum- until adult sizes in these taxa. Our color conver- firmatus cavespiram grew more whorls in order gences are fortuitous results of the influence of to attain a high spire, then the convergence is developmental heterochrony upon a universal adaptive since high spires were selected directly ontogeny of color. in both taxa. If the high spire of P. (G.) cir- Case 1: Because they are small, shells of P. cumfirmatus cavespiram is a by-product of (Gastrelasmus) rarely advance beyond the strong allometry prolonged to an advanta- flame stage to the formation of bands. P. (G.) geously large number of whorls, then the con- acutissimrus reaches its abnormally large size vergence is fortuitous. Again, I can neither dis- not by growing more whorls (with consequent tinguish these alternatives nor assess a third- intensification of development), but by enlarg- that both increased whorl number and a high ing the size of late whorls. All three subgenera dome were favored by selection, and that the of Poecilozonites have approximately the same mode of growth that links the two provided an number of adult postprotoconch whorls (5-6). easy mechanism for their mutual production. Color differences among subgenera are correlated with size-the two small subgenera rarely 4. Color and thickness form bands, but only the developmentally-re- Although band numbers, widths, positions, tarded P. (Poecilozonites) lack them. However, and times of formation differ within and among within any subgenus, given stages in the ontog- taxa, the basic ontogeny of coloration is com- eny of color tend to occur at the same whorl mon to the entire genus. Deposition brings on number regardless of size at that whorl. Since the last half whorl of the protoconch as even- P. (G.) acutissimus does not have an unusually sided radial stripes. These continue for about large number of whorls, it remains in the flame one whorl of accretionary growth, but are grad- stage as an adult. In addition, there is a perva- ually transformed to the characteristic zigzags sive, though unexplained, correlation between (also called flames or blotches) that give the high spires and strong coloration in Poecilozon- genus its name (Poecilozonites = variegated- ites (Gould, 1969, p. 441 and p. 501). As the rel- banded). Next, constrictions form in those parts atively widest taxon of its subgenus, we might of the zigzag that will not form bands in the expect P. (G.) acutissimus to retain weak color adult. Finally, the remaining parts of the zigzag even if it possessed more than the usual number strengthen and widen, interstitial color is depos- of whorls. ited between zigzags and bands form. The se- The retention of flames in adult P. (G.) acu- quence of flame-band transitions is also invari- tissimus is a consequence of normal growth pat- ant, moving from bottom to top of the whorl. terns for the subgenus. P. (P.) bermudensis The subperipheral color band always forms paedomorphs, on the other hand, form no bands first, followed by the supraperipheral (if pres- because their developmental rates are retarded ent), and finally by the subsutural (if present). relative to size. The adaptive significance of The subsutural band forms so late in ontogeny, this retardation lies in the thin shell it produces that most taxa do not produce it at all, but in calcium-poor habitats; weak coloration is an maintain color blotches in this position as accidental by-product of an evolutionary mecha- adults. There is enormous variation in whorl nism utilized to produce a thin shell rapidly. In number of the flame-band transition among neither case does the retention of flames have taxa. The subperipheral band, for example, may adaptive significance for its own sake. The form by the second postprotoconch whorl (most same result is reached in a different way in each P. (P.) cupula subspecies) or still be repre- case. The resemblance is fortuitous.

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions CONVERGENCE IN BERMUDIAN LAND SNAILS 415

Case 2: If P. (P.) cupula daUi had become have had the same significance that it did in P. small by reducing its number of whorls, shell (P.) bermudensis paedomorphs. P. (P.) cupula coloration would not have progressed beyond dalli retains its relatively thick shell because it the flame stage. But since miniaturization of decreased in size by miniaturizing its whorls whorls formed its path to small size, P. (P.) cu- without altering its rate of development. P. pula dalli retains the banded adult coloration of (G.) circumfirmatus cavespiram has an unusu- its ancestors. Its color does not correspond to ally thick shell for its subgenus because it grew the normal, flammulated pattern of P. (Gastre- more whorls and thereby prolonged its period of lasmus). But P. (G.) circumfirmatus cavespi- development to a stage of shell thickness at- ram does develop bands for two reasons: 1) It tained by no other taxon of its subgenus. I have has an unusually large number of postproto- no reason to believe that retention or acquisition conch whorls (6-7) and thereby reaches stages of a thick shell played a direct part in the adap- of the universal color ontogeny that normal tation of these forms. taxa of its subgenus do not attain. 2) It par- A CONCLUSION AND MODEST PROPOSAL takes of the common correlation between high spires and strong coloration. The preceding examples remain trivial in In both taxa, color bands are developed as a themselves and even a bit whimsical for their simple consequence of growth. I have no evi- fortuitous precision. Yet, I detect a larger sig- dence that these modes of growth were selected nificance for these cases because they exemplify, to produce a favored color rather than for their so very well, two general aspects of conver- own sake and suspect, therefore, that similari- gence that rarely receive the emphasis they de- ties in color are fortuitous. serve. Other allometric features of growth are com- 1. Forms of apparent complexity are often mon throughout the genus; they produce addi- generated by few factors. We doubt (or wonder tional fortuitous convergences that are visually inordinately at) many precise convergences be- less striking than that involving color: cause we view each of their attributes as a sepa- 1. umbilical width. In Poecilozonites, umbili- rate event. Yet, as Raup has shown in his com- cal width increases regularly until the shell puter simulations (1966, 1968, 1969), the inter- reaches its maximum relative width; thereafter action of simple rates and gradients upon initial it remains constant (Gould, 1969, fig. 6, p. 432). shapes during growth can produce such remark- The few relatively high-spired taxa of Poecilo- ably intricate structures as the echinoid ambu- zonites are exceptional. In these, umbilical lacrum. In most fields that deal with complex width, after reaching its maximum in the nor- situations, progress from near-blind empiricism mal way, becomes progressively constricted as to general theory comes when that complexity is rapidly increasing height/width ratios form a displayed as a result of few causal factors. (We high dome (Gould, 1969, photo on frontispiece). are, I believe, not witnessing this progression In accordance with these patterns, both P. (G.) in the ideas of mathematical population ecology acutissimus and P. (P.) bermudensis paedo- -what Wilson (1969) has so aptly termed morphs have equally wide umbilici, while both "post-Darwinism"). Morphology, another field P. (G.) circumfirmatus cavespiram and P. (P.) that deals in complexity, has long been ham- cupula dalli have narrowly constricted ones. pered by a tradition that exalts detailed, part- These resemblances are mechanical conse- by-part description as virtuous because it is "pure quences of similar shell shapes; we need not fact" and not theory-loaded-without realizing postulate any direct selection for their presence. that such a procedure subtly nudges us to a the- 2. shell thickness. Relative shell thickness in- ory that complexity is irreducible. I have argued creases constantly during the ontogeny of all elsewhere (chapter 3 of Gould, 1970) that the Poecilozonites, but the rate of increase varies reduction of complexity to fewer generating greatly. The thin shells of P. (P.) bermudensis factors would provide an essential component to paedomorphs are produced by relative retarda- a causal morphology. tion of the developmental rate; this, as I argued We meet, in our convergences, two types of before, is an adaptation to the low lime content character-groupings that can be reduced to sim- of Bermuda's red soils. P. (G.) acutissimus re- pler casual factors: tains, at its increased size, the thin shell that a) mechanical correlates. It has long been characterizes its subgenus because it reaches recognized that the exuberant diversity of mol- this size at a normal number of whorls and luscan shells can be rendered by varying the shows no sign of any prolongation of develop- few factors that must be specified to generate a ment. Since P. (G.) acutissimus lived only in spiral. D'Arcy Thompson (1942, p. 785) notes carbonate dunes, the thin shell is not likely to that Swammerdam had stated this as early as

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions 416 STEPHEN JAY GOULD

1737: "Omnis enim quae inter eas aznimadverti- coiled in large forms (Gould, 1966b, p. 591). tur differentia ex sola nascitur diversitate gyra- Several authors have documented similar fea- tionum"-for all difference that is noted among tures evolved independently by many larger for- them originates only from the diversity of spi- aminifers to meet structural problems of their rals. Since we are used to expressing our obser- size (Dunbar, 1963 for fusulines; Drooger, 1956 vations in widths, lengths and ratios (the prod- and Rat, 1963 for orbitoids). And a major event ucts, not the causes), we rarely apprehend the in the history of life, the evolution of internal redundancy. In this study, for example, conver- organs as concentrations of surfaces (lungs, in- gence in umbilical width is only the necessary testine, circulatory tubes), was so necessary for consequence of convergence in spire shape. the existence of large organisms rounder than a "Even strong and apparently complex resem- tapeworm that the presence of similar struc- blances in form . . . may sometimes signify no tures need not imply their presence in a common more than the... numerical coincidences which ancestor. The precision of size-imposed conver- are manifested in identity of length or weight gence is often striking. Andrewsarchus, the or any other simple magnitude" (Thompson, enormous terminal member of mesonychid lin- 1942, p. 808). eage that passed from cursorial carnivorous to b) developmental correlates. In land snails, carrion feeding to omnivorous life at increased the changes that mark adulthood are widespread sizes, evolved a remarkable likeness to the later and varied. Land snails often change color, shell omnivorous entelodonts (Szalay and Gould, thickness and coiling direction at this stage and 1966). these features arise in concert, presumably un- Both convergences reported in this paper der the influence of a single sex or growth hor- were initiated by the entrance of one taxon into mone. In Poecilozonites, development is often the size range of another subgenus. In one case, accelerated or retarded relative to the rate of the new and larger size required an altered size increase. In paedomorphosis, characters of shape that ensured convergence upon a taxon of spire shape, color and thickness are delayed to the large subgenus. The mechanical correlates an equal degree in size of appearance (Gould, of these size-required changes then intensified 1968). Their correlated modification probably the extent of an already precise convergence. marks a change in a single factor that monitors If the principles embodied in my two state- the rate of development. ments were more widely applied, we would have 2. Change in size requires correlated modifi- to accept the unpleasant truth that convergence cations of shape; these are often so rigidly spec- is a more fundamental phenomenon than we ified that convergence is inevitable. have allowed. I label this unpleasant for two With these two statements, I am arguing that reasons. First, it confounds the efforts of taxon- close convergence is less surprising and improb- omists working on fossils with simple hard parts able than it appears under our usual views of (where the tired argument that all conver- phylogeny. The first statement, discussed above, gences can be recognized if enough characters argues that the mechanism of a precise conver- are studied does not apply). The enormous tax- gence in many characters may be simple because onomic frustrations of homeomorphy have been it involves the modification of a single factor well expressed by Ager (1965) for Jurassic with many correlated effects. This second state- brachiopods. But I think we have finally ment contends that a close convergence may be reached an age in which increased difficulties in inevitable because it represents the only ordering will no longer argue against a biologi- adaptive solution to a problem faced in common cal principle. Secondly, convergence does not fit by several lineages. In this realm of biomechan- well on the traditional tree that we choose for ics, no factor imposes more serious constraints our phylogenetic representations. Most of the upon form than large changes in size itself. quantitative phyletics of modern numerical tax- Best documented are the relative increases in onomy presuppose a continuing divergence, surface area that large animals require to main- while other recommendations for a phylogenetic tain the processes of respiration, digestion and classification (Hennig, 1966) would consider support; the geometric pathways to this in- only the branching points of a diverging system. crease-flattening, elongation and convolution- What we need, I expect, are new pictures of are strictly limited (Thompson, 1942; Gould, phylogeny (strawberry runners or combs), and 1966b; Alexander, 1968). The fossil record pro- perhaps a new vocabulary like that proposed by vides many examples of size-imposed similari- Huxley (1958). ties among lineages increasing in size. The The case of convergence will, perhaps, best be brachiopod lophophore gathers food through its made not by a theoretical assault, but by the surfaces and must be multi-lobed or spirally documentation of such harmless, trivial, and

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions CONVERGENCE IN BERMUDIAN LAND SNAILS 417 whimsical examples that can propel this un- ram = beware the spire) is given to warn of pleasant truth through a back door of our evolu- possible confusion between this taxon and the tionary literature. For as Gilbert's jester, Jack more common, equally small and high spired P. Point, once remarked: cupula dalli. MCZ29021 When they're offered to the world in merry guise, Repository.-Holotype (Inverte- Unpleasant truths are swallowed with a will- brate Paleontology: Gastropoda), Museum of For he who's make his fellow-creatureswise Comparative Zoology, Harvard University, Should always gild the philosophicpill! Cambridge, Massachusetts. SYSTEMATICPALEONTOLOGY ACKNOWLEDGMENTS ZONITIDAE Family Supported in part by NSF Grant GA-901. Subfamily POECILOZONITINAE This is a contribution from the Bermuda 1924 paper Pilsbry, Biological Station for Research. I thank C. Genus POECILOZONITES0. 1884 Jones, Boettger, H. Holland, and R. for and POECILOZONITES Adlington figures Subgenus (GAsTRELASMUS) photography. Pilsbry, 1924 Species P. CIRCUMFIRMATUS (Redfield) REFERENCES Ager, D. V. 1965. The adaptations of Mesozoic Poecilozonites (Gastrelasmus) circumfirmatus brachiopodsto different environments.Palaeogeog. cavespiram, n. subsp. Palaeochim. Palaeoecol. 1:143-172. R. McN. Text-figure 4 Alexander, 1968. Animal Mechanics. Univ. of Washington Press, Seattle. 346 p. Description.-Adult with 5y2-7 postproto- Bretz, J. H. 1960. Bermuda: a partially drowned,late conch whorls in is 9.8 mature, Pleistocene karst: Bull. Geol. Soc. Am., v. (6 holotype); holotype 1729-1754. in 8.6 in tall and in- 71, p. mm width, height; spire Dean, B. 1908. Accidental resemblancesamong ani- tensely domed; whorl periphery only slightly mals. A chapter in un-natural history. Pop. Sci. angulate; aperture rounded; umbilicus con- Monthly 72:304-312. C. W. 1956. Parallel in stricted; lamella terminating just Drooger, evolutionarytrends strong palatal larger Foraminifera. Proc. Koninkl. Nederl. Akad. inside aperture at whorl periphery; subperi- Weten. 59:458-469. pheral color band present in adult; flames still Dunbar, C. 1963. Trends of evolution in American present in subsutural band position; supraperi- Fusulines. In G.H.R. von Koenigswald et al. trends in Foraminifera. Else- band sometimes present, sometimes as (eds.), Evolutionary pheral vier, Amsterdam. flames. George, T. N. 1962. The concept of homeomorphy. Comparison.-Highest spired, most intensely Proc. Geologists' Assoc. 73:9-64. domed, most narrowly umbilicate and most Gould, S. J. 1966a. Allometry in Pleistocene land colored taxon of its all these snails from Bermuda: The influence of size upon strongly subgenus: 40:1131-1141. the shape. J. Paleontology features are correlated in development or . 1966b.Allometry and size in ontogeny and phy- mechanics of spiral growth. logeny. Biol. Rev. 41:587-640. Occurrence.-Shore Hills Soil, western part -. 1968. Ontogeny and the explanation of form: an allometric In D. B. Macurda of St. George's Island, Bermuda. Locality 48 of analysis. (ed.), Paleobiologicalaspects of growth and development, Gould, 1969. More than 50 specimens found a symposium. Paleont. Soc. Mem. 2, J. Paleontol- with P. (P.) cupula cupuloides. ogy 42 (5, suppl.) :81-98. Etymology.-The subspecies name (cavespi- . 1969. An evolutionary microcosm: Pleistocene and Recent history of the land snail P. (Poecilo- zonites) in Bermuda. Bull. Mus. Comp. Zool. 138: 407-532. . 1970. Evolutionary paleontologyand the science of form. Earth-Sci. Rev., in press. Haas, 0., and G. G. Simpson. 1946. Analysis of some phylogenetic terms. Proc. Amer. Phil. Soc. 90: 319-347. Hennig, W. 1966. Phylogenetic systematics: Univ. of Illinois Press, Urbana, Ill. 266 p. Huxley, J. S. 1958. Evolutionary processes and tax- onomy with special reference to grades. Uppsala Univ. Arsskr., p. 21-38. Land, L. S., F. T. Mackenzie, and S. J. Gould. 1967. Pleistocene history of Bermuda. Bull. Geol. Soc. Amer. 78:993-1006. Pilsbry, H. A. 1924. Recent and fossil Bermudan (sic) snails of the genus Poecilozonites. Proc. Acad. Nat. Sci. Philadelphia76:1-9. TEXT-FIG. 4-Holotype of P.(G.) circumfirmatus Rat, P. 1963. L'accroissementde taille et les modifi- cavespiram.Actual width of specimenis 9.8 mm. cations architecturalescorrelatives chez les orbito-

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions 418 STEPHEN JAY GOULD

lines. In G.H.R. von Koenigswald et al. (eds.), Stevens, S. S. 1968. Measurement,statistics and the Evolutionary trends in Foraminifera. Elsevier, schemapiricview. Science 161: 849-856. Amsterdam.p. 93-109. Stevens, S. S. and M. Girao. 1963. Subjective scaling Raup, D. M. 1966. Geometric analysis of shell coil- of length and area and the matching of length to ing: general problems. J. Paleontology 40: loudness and brightness. J. Exp. Psychol. 66: 1178-1190. 177-186. 1968. Theoretical morphology of echinoid Szalay, F. S., and S. J. Gould. 1966. Asiatic Mesony- growth. In Paleobiological aspects of growth and chidae (Mammalia, Condylarthra). Bull. Amer. development,a symposium: Paleont. Soc. Mem. 2, Mus. Nat. Hist. 132:127-174. J. Paleontology 42 (5, suppl.) :50-63. Thompson, D'Arcy W. 1942. On growth and form. Raup, D. M., and A. Seilacher. 1969. Fossil foraging CambridgeUniv. Press, Cambridge.1116 p. behavior: computer simulation. Science 166: Wilson, E. 0. 1969. The new populationbiology. Sci- 994-995. ence 163:1184-1185. Rudwick, M. J. S. 1965. Adaptive homeomorphyin the brachiopodsTetractinella and Cheirothyris.Pa- laont. Zeitschr.39:134-146. MANUSCRIPT RECEIVEDMARCH 31, 1970

This content downloaded from 134.84.3.217 on Wed, 01 Apr 2015 22:46:03 UTC All use subject to JSTOR Terms and Conditions