Proc. Natl. Acad. Sci. USA Vol. 76, No. 8, pp. 4131-4135, August 1979 Neurobiology

Decussation geometries in the goldfish nervous system: Correlation with probability of survival (brain asymmetries/lateralization of function/Mauthner's neuron/optic chiasm/natural selection) RICHARD L. ROTH Department of Biological Sciences, Stanford University, Stanford, California 94305 Communicated by Donald Kennedy, May 9, 1979

ABSTRACT In the goldfish, the optic nerve decussation assumed to be representative of goldfish available in the United occurs without intermingling of fibers from the two eyes. In States. The embryonic and larval specimens examined were two-thirds of juvenile and adult specimens, the left optic nerve derived from two lots of fertile eggs and, by the breeders' es- is dorsal at the midline. In about 60% of the specimens, the decussation of Mauthner's neuron also has a left-dorsal-to-right timates, represent the progeny of 9-14 spawning females.* (L/R) configuration. Concordance for decussation geometry is Despite this limited sampling of embryos and larvae, the results greater than 80%, with smaller specimens accounting for a presented below indicate that they, too, provide an adequate disproportionate number of discordant cases. In embryos and basis for generalization. very young larvae, the L/R configuration occurs in slightly less For minimization of sampling bias, most of the Spanish moss than 50% of o tic chiasmata and in slightly more than 50% of into each Mauthner's cell chiasmata, and there is no significant tendency containing the eggs was teased and cut portions, toward concordance. However, larval specimens that survive bearing 50-75 eggs. These portions were distributed sequen- 1 month after hatching are markedly skewed toward adult de- tially among 10 aquaria so that each aquarium contained every cussation patterns and somewhat skewed toward concordance. tenth sample of eggs. From the remainder of the Spanish moss, These observations, together with results of cannibalization and individual eggs were isolated, and these singlets were distrib- predation experiments, suggest that the adult population pattern uted among trios of 1-gallon aquaria so that, as defined by the can be achieved through selection pressures against discordant individuals and, perhaps separately, against individuals with order of selection, one aquarium contained specimens 1, 4 ..., right-dorsal-to-left (R/L) chiasmatic configurations. Decussation n -2; a second contained specimens 2, 5, .. -,n-1; and a third patterns correlate with embryonic postures, larval orientation contained specimens 3, 6, . . . n. Some of these randomized preferences, and growth rates after hatching, which may samples were killed in their entirety at 1 day to 1 week after themselves serve as a basis for selection. hatching and used to establish a larval population base line with respect to optic and M-cell chiasmatic geometries. Others were Recent studies of lateral asymmetries in nervous systems have subsampled in a variety of ways, which are detailed below. been directed not only toward further cataloging of functional Maintenance of Embryos and Larvae. Eggs were main- asymmetries, but also toward examining the morphological tained in water treated with methylene blue, Fungizone, or a (1-5) and biochemical (6) asymmetries that may underlie la- Fungizone/penicillin mixture (GIBCO) to retard fungal and teralization of function. Among the efforts to define anatomical fed either asymmetries have been attempts to correlate geometric con- bacterial infections. Free-swimming larvae were figuration of the pyramidal tract decussation with "handedness" infusoria or sieved hard-boiled egg yolk, and water was changed (2) and to define a genetic basis for different decussation as necessary. geometries of the optic nerves and Mauthner cell (M-cell) axons Histological Procedures and Determination of Chiasmatic in teleost fish (7). These studies of asymmetry in decussation Configurations. Virtually all the juvenile and adult specimens patterning have produced equivocal results because functional on which this report is based had initially been used in other correlations could not be tested (5) 6r because of small or po- investigations in which it became an incidental practice to tentially biased samples (2, 7). record the configurations of optic chiasmata as brains were The present investigation of asymmetries in decussation being removed for histological processing. This earlier work geometries exhibited by the M-cells and optic nerves of the yielded a large library of Bodian-stained serial sections. Al- goldfish overcomes some of these problems by use of large though the M-cell was not included in sections of most of these samples, by following randomized samples from embryonic specimens and the optic chiasm was not included in others, the through late larval stages, and by sorting previously randomized configuration of the M-cell chiasm could be determined in more to a of and behavioral than 100 individuals and that of both optic and M-cell chi- samples according variety postural asmata in more than 80. The sample used to determine optic asymmetries. chiasmatic pattern included, in addition to specimens in this MATERIALS AND METHODS slide collection, approximately 100 poorly fixed specimens in which the optic chiasm was scored only by dissection. Structure of Samples. This report is based on the examina- Preparation of all larval specimens was according to the tion of more than 250 juvenile or adult and more than 1000 standard Bodian method. To facilitate accuracy in scoring of embryonic or larval goldfish (Carassius auratus). Because the juvenile and adult specimens were obtained over the past 14 Abbreviations: M-cell, Mauthner's neuron; L/R, left-dorsal-to-right; years from several distributors in Ohio and California, they are R/L, right-dorsal-to-left. * One lot of eggs, representing "domestic" stock and provided by Ozark The publication costs of this article were defrayed in part by page Fisheries, Inc. (Stoutland, MO), was produced by 8-12 females. The charge payment. This article must therefore be hereby marked "ad- other, representing recently imported Japanese stock and provided vertisement" in accordance with 18 U. S. C. §1734 solely to indicate by Golden State Hatcheries, Inc. (Winters, CA), was produced by this fact. one or two females. 4131 Downloaded by guest on September 27, 2021 4132 Neurobiology: Roth Proc. Natl. Acad. Sci. USA 76 (1979)

decussation patterns, conventions were uniform with respect Table 2. Preponderance of concordant chiasmatic configurations to orientation of specimens in the paraffin blocks, orientation in a sample of 81 juvenile and adult goldfish of paraffin ribbons on slides, and orientation of finished slides M-cell chiasm on the microscope stage. Each specimen was scored first with Optic L/R R/L respect to its apparent orientation in the microscopic field, then chiasm Obs. Exp. Obs. Exp. Total with respect to the apparent relationship between chiasmatic L/R n 42 29 8 21 components. Only after a permanent record of these apparent 50 f 0.518 0.358 0.099 0.259 0.617 relationships had been made were apparent relationships converted to the actual ones by defining the axes of the indi- R/L n 5 18 26 13 31 vidual specimen. f 0.062 0.222 0.321 0.161 0.383 RESULTS Total n 47 34 81 f 0.580 0.420 The optic and M-cell chiasmatic configurations in the sample Heavy lines indicate division of the sample into the four compart- of juvenile and adult specimens are summarized in Table 1. ments jointly defined by the geometries of the optic and M-cell chi- Because this sample was drawn about equally from midwestern asmata. Concordant cases appear in the upper left and lower right and west coast suppliers and because the frequencies of the two compartments. Each compartment shows the number of cases ex- chiasmatic geometries were nearly identical for the two sub- pected if chiasmatic patterns are independent and the number ob- samples, the data were pooled. The critical points are (i) the served. n, Number; f, frequency; Obs., observed; Exp., expected. population frequency of left-dorsal-to-right Differences between observed and expected values are highly sig- (L/R) configura- nificant (X2 = 36.19, P << 0.001). tions must be very nearly 0.67 for the optic chiasm and 0.59 for the M-cell chiasm, and (ii) the differences in frequency of L/R configurations of the two chiasmata are significant. Restructuring of larval populations is evident within a few Considered separately, these asymmetries in pattern might weeks of hatching. Table 4 summarizes the distribution of be interpreted as trivial accidents of development. However, chiasmatic configurations in 3- to-6-week-old survivors of ini- those specimens in which both chiasmata could be examined tially randomized samples (matched to those killed at 1-7 days suggested that patterning of decussations might be of consid- after hatching). In this older population the asymmetric dis- erable importance. As shown in Table 2, the configurations of tribution of chiasmatic geometries approaches that of the ju- optic and M-cell chiasmata are neither independent of each venile/adult sample. This population restructuring is evidently other (in which case concordance for pattern would be about the result of selective cannibalization among larvae of the same 50%) nor are they rigorously predisposed to a common pattern chronological age rather than of differential low resistance to (in which case concordance would be perfect). One might still infection or of postembryonic expression of lethal genes or se- argue that the imperfect (80%) concordance results from partial rious developmental anomalies. Those recently dead or mori- dependence on some common event or condition during em- bund specimens in which decussation patterns could be de- bryogenesis were it not for the observation that discordant termined represented all four population compartments in the patterns were more common in the smaller members of this proportions expected on the basis of chronological age. Fur- sample. This feature of the juvenile/adult sample suggested an thermore, dead and dying specimens accounted for only a small easily testable alternative hypothesis-namely, that develop- fraction of the observed attrition in larval populations. mental forces might yield populations randomized with respect to decussation geometries and that the adult population dis- Cannibalization/predation experiments tribution might be a consequence of differential survival. This differential survival hypothesis is supported by the Two trios of matched samples of free-swimming larvae (3 weeks population distribution of optic and M-cell chiasmatic patterns after hatching) were used in pilot experiments designed to test in a large randomized sample of 1- to-7-day-old larvae (Table whether the acute selection pressure provided by a hungry, 3). Comparison of this young larval sample with the juvenile/ larger fish is uniform or differential in terms of chiasmatic adult sample (Table 2) reveals two differences and one simi- geometries of prospective prey. From initially randomized larity: (i) The distribution of larvae among the four population samples, larvae were drawn seriatim and placed sequentially compartments is almost exactly predicted by the assumption in three aquaria in the way used for randomization of eggs. One that the decussation geometries of optic nerves and M-cell axons member of each trio of matched samples was fixed immediately are determined independently. (ii) The larval sample slightly and served as a control population, while the other two were favors the right-dorsal-to-left (R/L) configuration of the optic exposed to a predator (a large guppy, swordtail, or black tetra) chiasm, which is characteristic of somewhat fewer than one- or cannibal (a larger, so-called "feeder" goldfish). After the third of the adult specimens. (iii) The frequency of L/R M-cell population in an experimental pool had been reduced by about chiasmata is only slightly lower in the larval than in the adult one-half, which required between 2 and 30 min, all surviving sample. larvae were fixed. The cumulative results of the four separate experimental trials (Table 5) show that selection of prey is sig- the structure of Table 1. Incidence of L/R configurations of optic and M-cell nificantly nonrandom. However, population chiasmata in juvenile and adult the survivors deviates from expectations based on the juve- goldfish nile/adult sample in that there are proportionately fewer R/L 95% con- concordant specimens and proportionately more specimens No. of Fre- fidence with an L/R optic chiasm and an R/L M-cell chiasm. As will L/R quency limits for be argued below, these differences probably represent artifacts Sample configu- of L/R observed of the experimental conditions. There was no significant size Chiasm size rations pattern frequencies difference between the control and the experimental popula- Optic nerve 256 172 0.67 0.610-0.726 tions. Thus, it is clear that selection under these extreme con- M-cell axon 115 68 0.59 0.495-0.680 ditions did not particularly disfavor smaller individuals. Downloaded by guest on September 27, 2021 Neurobiology: Roth Proc. Natl. Acad. Sci. USA 76 (1979) 4133 Table 3. Independence of chiasmatictonfiguations in two samples of 1-day-old to 1-week-old larvae M-cell chiasm L/R R/L Optic Sample A SampleB A+B Sample A SampleB A+ B chiasm Obs. Exp. Obs. Exp. Obs. Exp. ObIs. Exp. Obs. Exp. Obs. Exp. Total L/R n 61 59 71 70 132 129 48 50 53 54 101 104 233 f 0.272 0.263 0.284 0.282 0.278 0.273 0.2i4 0.224 0.212 0.214 0.213 0.219 0.492 R/L n .60 62 71 72 131 134 5$ 53 55 54 110 107 241 f 0.268 0.277 0.284 0.286 0.276 0.281 0.246 0.236 0.220 0.218 0.232 0.226 0.508 Total n 263 221 474 f 0.555 0.445 Abbreviations are as for Table 2. Sample A represents progency of 8-12 domestic females; sample B, progeny of 1-2 recently imported females. Frequencies of M-cell configurations deviate significantly from equality (X2 _ 3.9325, P < 0.05). Differences between observed and expected values in the four population compartments are not statistically significant (X2 = 0.2532, P - 0.60). Behavioral and developmental correlates of the head and tail deviate from the projected midsagittal plane decussation pattern of the trunk. Such postural asymmetries persist in dechorionated embryos and so are not simply responses to physical constraint. Preliminary work has revealed a number of behavioral and Individual embryos exhibit preferred postures-e.g., the tip developmental correlates of decussation pattern. Each of these of the tail projects either to the right or to the left of the midline correlates is based on, a sample of more than 100 individuals of the head. Such "tail-right" and "tail-left" postures signify and, except for the orientation preference result, which is sig- both neuroanatomical and behavioral differences. Tail-right nificant at the 10% level of confidence, each is statistically postures correlate with concordant chiasmatic patterns and with significant at a 5% or 1% level. These results will be documented eventual strong larval orientation preferences; tail-left postures elsewhere but, because they provide a reasonable basis for correlate with discordant chiasmatic patterns and with weak differential survival, a brief description is given here. Orientation and Depth Preferences. Individuals with dis- orientation preferences. At on cordant patterning of optic and M-cell'chiasmata swim about Growth Rate. hatching, concordant larvae are, the equally in clockwise and counterclockwise directions. Con- average, 10% smaller than discordant ones; 3 days after feeding cordant individuals have strong preferences to swim in either begins, concordant larvae are 5% larger. Thus, as a group, grow clockwise or counterclockwise directions, although neither concordant larvae faster than their discordant siblings. orientation preference is correlated with a particular chiasmatic Lateral Size Asymmetries in the Nervous System. The configuration. Chiasmatic patterns also'appear to correlate with configuration of the M-cell chiasm is correlated with a slight preferred swimming depth. For three samples of free- (2-5%) lateral asymmetry in size of the medulla. Typically, the swimming larvae (totaling about 300), depth of capture was M-cell of the locally larger side of the medulla contributes the monitored. Within each of these samples, larvae caught near ventral component of the chiasm. Also, the ventral component the top of the aquarium had a disproportion of concordant L/R of the optic chiasm generally originates in the larger eye. For chiasmata; those caught near the bottom had a disproportion both decussations there are distinctive differences in the pitch of concordant R/L chiasmata. Although such evidence of of prechiasmatic portions of the dorsal and ventral components, preferred swimming depths was less clear for discordant larvae, but it is not yet clear whether these differences in pitch are a those with L/R-M-cell chiasmata tended to be caught near the cause or an effect of decussation geometry. top of an aquarium; those with R/L M-cell chiasmata, near the Egg Size. Egg size per se appears to influence decussation bottom. Thus, the geometry of the M-cell chiasm alone is a patterning. Larger eggs yield a disproportionate number of significant correlate of depth preference. larvae with R/L optic chiasmata, the overproduction being Embryonic Posture. Even in early tailbud stages the goldfish principally of discordant individuals with L/R M-cell chi- embryo is wrapped asymmetrically about the yolk mass so that asmata.

Table 4. Chiasmatic configurations in 3- to 6-week-old survivors from initially randomized samples M-cell chiasm Optic L/R R/L chiasm Obs. Exp. Exp.* Obs. Exp. Exp.* Total L/R n 57 46 53 35 36 39 92 f 0.341 0.278 0.320 0.210 0.213 0.231 0.551 R/L n 40 46 44 35 39 31 75 f 0.239 0.276 0.261 0.210 0.232 0.188 0.449 Total n 97 70 167 f 0.581 0.419 The table makes two points: that the optic chiasm configuration has shifted toward the adult frequency and that a trend toward concordance is developing. Exp., expected from observed frequencies in Table 3; Exp.*, expected if chiasmatic patterns of this sample are independent; other abbreviations are as for Table 2. Deviations from expectation based on Table 3 are significant (X2 = 3.5618, P - 0.05; with Yates' correction X = 3.308, 0.05 < P < 0.10). Statistical significance of the trend toward concordance fails to meet the 5% confidence level (X2 = 1.2619, P 0.25; with Yates' correction x2 = 0.9326, P - 0.30). Downloaded by guest on September 27, 2021 4134 Neurobiology: Roth Proc. Natl. Acad. Sci. USA 76 (1979) Table 5. Chiasmatic configurations in matched samples used in acute predation experiments M-cell chiasm Control Experimental Optic L/R R/L L/R R/L chiasm Obs. Exp. Obs. Exp. Total Obs. Exp. Exp.* Obs. Exp. Exp.* Total L/R n 14 13 8 9 22 15 12 15 11 7 11 26 f 0.341 0.316 0.195 0.224 0.540 0.441 0.341 0.427 0.324 0.195 0.337 0.765 R/L n 10 11 9 8 19 4 8 4 4 7 4 8 f 0.244 0.269 0.220 0.191 0.460 0.118 0.244 0.131 0.118 0.220 0.104 0.235 Total n 24 17 41 19 15 34 f 0.585 0.415 0.559 0.441 The control sample combined two samples, one consisting of 20, the other of 21 individuals. The experimental sample consisted of survivors of four separate predation episodes, each ofwhich involved starting samples of 21 larvae. For the control sample, abbreviations are as for Table 2. For the experimental sample: Exp., expected from observed frequencies in control sample; Exp.*, expected ifchiasmatic patterns ofthis sample are independent; other abbreviations as for Table 2. There is no significant trend toward concordant patterning in either sample. The differences between control and experimental samples are significant (X2 = 7.6924, P < 0.01; with Yates' correction x2 = 8.1278, P < 0.005). DISCUSSION Tonai, personal communication). The faster growth rate of concordant individuals would be expected to favor selection The Selection Phenomenon. The fundamental conclusion against discordance simply because the generally larger con- to be drawn from the above data is that different configurations cordant individuals are more likely to eat than to be eaten. of optic and M-cell chiasmata are highly correlated with dif- The differential growth phenomenon may be mediated, in ferent survival probabilities. Two observations suggest that an part, by species-specific "crowding factors" (8, 9). Because L/R optic chiasm per se confers a selective advantage. (i) Whereas the population frequency of L/R M-cell chiasmata fertile eggs are hatched in "fresh" water, the titer of such remains virtually constant, that of L/R optic chiasmata shifts growth inhibitors probably does not reach significant levels until discordant several days after hatching. By this time concordant fry would from about 0.50 to nearly 0.70. (ii) Considering only already be larger, and growth inhibition would preferentially individuals, those with L/R optic chiasmata are about twice as discordant effects and early differential likely to survive. The data on pond-reared juveniles and adults affect fry. Crowding (Table 2) indicate that concordant decussation patterning is growth call into question the common assumption that size and of 3- to 6-week-old age are directly correlated. Although the principal conclusions particularly advantageous. Comparison it be necessary to larvae (Table 4) with younger larvae (Table 3) favors this of this report would not be affected, may conclusion. However, the results of the acute cannibalization reinterpret the very finding on which the study of larvae was experiments (Table 5) show neither the sparing of concordant predicted. Specifically, the smaller average size of discordant members of the juvenile/adult sample may have been a con- R/L individuals nor the deletion of discordant individuals with R/L M-cell chiasmata predicted from the juvenile/adult sequence of discordance rather than of a shorter exposure to selection pressures. sample. is It is unlikely that sampling bias can account for these dif- From the acute cannibalization experiments it apparent in the outcome of "natural" and laboratory selection. that size alone does not account for selection. The preliminary ferences intuitive ex- The more likely explanation lies in several marked differences behavioral-anatomical correlations confirm the between conditions in commercial rearing ponds and in the planation of this result-namely, that there must be systematic laboratory. The population in a rearing pond consists of indi- behavioral differences. Direct observation and cinematographic that the and cannibals viduals no more than 2 or 3 days apart in age, and there is cover records indicate predators preferentially to which threatened fish can escape. In the acute cannibaliza- attacked deeper swimming larvae. Thus, themost obviously tion experiments the size disparity of cannibals and prospective relevant of the behavioral biases associated With decussation prey was at least 50 times that typical of a pond; no cover was pattern is depth preference. Whether orientation-preferences provided; and, because only one or two cannibals were used in also contribute to the selection process is not yet known. A each experiment, there was less opportunity for behavioral plausible but untested hypothesis is that individuals with strong biases of cannibal and prey to interact. orientation preferences have more efficacious escape mecha- Although further study is needed to establish causal rela- nisms or use different escape strategies. tionships, dimorphic nervous decussations appear to yield two Probable Involvement of Lateral Asymmetries. Although consequences: (i) the resultant variety of behavioral biases the configurations of neural decussations are almost certainly ensures coverage of available microniches and (ii) differential epiphenomena, the results of this study have, so far, been pre- growth rates ensure fodder for faster growing individuals. sented only in terms of decussation geometry. Admittedly, there The Selection Process. Population selection among goldfish is no conclusive evidence as to the nature of the actual causal probably has several determinants which assume different agencies; however, the developmental correlates of decussation priorities according to environmental conditions. An initial bias pattern implicate lateral asymmetries of the nervous system and toward the usual outcome is evident in both samples of young neuroeffector apparatus. In particular, the correlation of larvae-namely, there are slightly more concordant and slightly right-left size disparities and decussation pattern suggests that, fewer discordant individuals than predicted by perfect inde- via a functional superiority of the larger of sister structures, pendence of optic and M-cell chiasmatic pattern. Despite the something akin to "cerebral dominance" may be involved. It low level of statistical confidence in this finding, it cannot yet is postulated that embryonic postural preferences and the be- be excluded as a partial basis for population restructuring. havioral biases of larvae are also compatible with a scheme Under the usual rearing conditions, cannibalization among fry based on lateral asymmetries. Embryonic postures are pre- of the same age is the major source of population attrition (F. sumed to reflect the longitudinal distribution of asymmetries Downloaded by guest on September 27, 2021 Neurobiology: Roth Proc. Natl. Acad. Sci. USA 76 (1979) 4135 in the development of axial musculature. Behavioral biases are sign. Further neuroanatomical study is required to establish presumed to reflect a composite of neural and somatic asym- whether other expressions of neural asymmetry in flatfishes metries except that depth preferences may be attributable follow the same rigid pattern exhibited by the optic chiasm. At primarily to differences in swim bladder dynamics, which may this time, these results emphasize the possibility that population depend on systematic correlations between neural asymmetries selection in goldfish is related to interaction of somatic and and visceral disposition (10). neural asymmetries in producing the behavioral biases on which Evidence that right-left size disparities stem from asymmetric selection appears to depend. distribution of oocytic materials, recently reviewed by Morgan (11), dates back to observations of Larrabee (12) on naturally The technical assistance of M. Klein, J. Lindzey, G. Nobari, and S. Williams is gratefully acknowledged. I thank M. Feldman, C. S. occurring double embryos in trout and of Spemann and Goodman, D. Kennedy, S. R. Johnson, D. Perkel, T. L. Page, and J. J. Falkenberg (13) on experimentally produced double embryos Wine for discussions and critical reading of the manuscript. Particular in newts. The general proposition is that asymmetries in oocytic thanks are extended to M. Metzger and F. Tonai, who provided the factors result in slight inequalities in the division of the embryo embryos and readily shared their practical knowledge of goldfish into right and left sides. Whether eventual morphological breeding habits and natural history. This work was supported in part asymmetries result from unequal parceling of starting materials by Grant NS02944 from the National Institutes of Health, U.S. Public or from lateral asymmetries in DNA synthesis (14) or rate of Health Service. differentiation and growth is unresolved. Morgan's postulate (11) that differential development uniformly favors the left side 1. Braitenberg, V. & Kemali, M. (1970) J. Comp. Neurol. 138, 137-146. is not consistent with the present results. Rather, with reference 2. Kertesz, A. & Geschwind, N. (1971) Arch. Neurol. 24, 326- to the midsagittal plane, the zygote or early embryo must consist 332. of a longitudinal array of independent sectors in which asym- 3. Morgan, M. J., O'Donnell, J. & Oliver, R. F. (1973) J. Comp. metries in starting materials or developmental timetables may Neurol. 149, 203-214. favor either the right or left side. Furthermore, it appears that 4. Nottebohm, F. & Nottebohm, M. E. (1976) J. Comp. Physiol. A egg size itself modulates the distribution of asymmetries (ref. 108, 171-192. 12, unpublished observations, and the present investigation). 5. Yakovlev, P. I. & Rakic, P. (1966) Trans. Am. Neurol. Assoc. 91, Generality of Selection Phenomenon. Work in progress on 366-367. six other teleost species, representing four families, suggests the 6. Oke, A., Keller, R., Mefford, I. & Adams, R. N. (1978) Science generality, at least among teleosts, not only of differential 200, 1411-1413. of behavioral 7. Levinthal, F., Macagno, E. & Levinthal, C. (1975) Cold Spring survival, but also differential growth and biases Harbor Symp. Quant. Biol. 40, 321-331. as correlates of decussation patterning (unpublished observa- 8. Francis, A. A., Smith, F. & Pfuderer, P. (1974) Prog. Fish Cult. tions). It is also significant that almost 80 years ago Duncker (15) 36, 196-200. reported selection against sinistral individuals of the dimorphic 9. Yu, M.-L. & Perlmutter, A. (1970) Growth 34, 153-175. flounder, Pleuronectes flesus. Duncker's conclusion was based 10. Woellwarth, C. von (1950) Wilhelm Roux' Arch. Entwick- on relative proportions of dextral and sinistral individuals in lungsmech. Org. 144, 178-256. different size classes of a large sample. Thus, the results could 11. Morgan, M. J. (1977) in Lateralization in the Nervous System, reflect differential growth rather than predation or some other eds. Harnad, S., Doty, R. W., 'Goldstein, L., Jaynes, J. '& Krau- source of differential mortality. In either event, it is unlikely thamer, G. (Academic, New York), pp. 173-194. that body form itself accounts for the differences; rather, dif- 12. Larrabee, A. P. (1906) Proc. Am. Acad. Arts Sci. 42,215-231. ferences in somatic asymmetry must be associated with neu- 13. Spemann, H. & Falkenberg, H. (1919) Wilhelm Roux' Arch. Entwicklungsmech. Org. 45,371-422. rologic differences. Neuroanatomical work (16) has shown that 14. Vargas-Lizardi, P. & Lyser, K. M. (1974) Dev. Biol. 38, 220- in all individuals of generally dextral flatfishes, such as P. flesus, 228. the dorsal component of the optic chiasm originates in the left 15. Duncker, G. (1900) Wiss. Meeresuntersuchungen N. F. 3, eye. Selection against (or growth retardation of) sinistral indi- 333-406. viduals is, therefore, equivalent to selection against individuals 16. Parker, G. H. (1903) Bull. Mus. Comp. Zool. Harv. Coll. 40, in which neurologic and somatic asymmetries are of opposite 221-241. Downloaded by guest on September 27, 2021