The fossil records of Cryptopecten vesiculosus can be traced back to the Middle Pliocene (ca. 3.5 Ma), Though some dead-end incipient species may have arisen from this stock, many fossil samples seem to indicate the persistency and phyletic morphological change of this species. The geographic distribution of this pectinid, especially the northern and southern limits, must have fluctuated during the geologic past in accordance with changing marine conditions, but its center seems to have remained in the seas surrounding the Japanese Islands. So far as I am aware, there is no undoubted fossil occurrence of C. vesiculosus in countries other than Japan.

More than 10 Pliocene, more than 20 Pleistocene and one Holocene fossil samples in addition to a number of Recent ones are available for the study of phyletic evolution of this species (see List of Examined Samples (pp. 127-137) for their localities, horizons, etc.), though some of them are not very useful owing to the small sample size and ambiguous geologic age. Many characters of these fossil samples, except for some Late Pleistocene and Holocene ones, are clearly outside the range of geographic variation in the present seas, and are mainly attributable to evolutionary change. At the same time it is here recognized that some characters changed directionally with time.

1.Shell Size and Growth Rate [see Table 13]

The overall shell height of the largest individual () and the average shell height at the formation of the first stepwise growth ring () for several selected large fossil and Recent samples were compared. The value of is inevitably influenced by sample size, but in this case a large sample does not necessarily show a large value of . It may be concluded that the ultimate size of gerontic individuals generally decreases with time. Large individuals over 30 mm are commonly found in every Pliocene and Pleistocene sample, while even gerontic individuals with many growth rings seldom exceed 30 mm in Holocene and Recent samples.

The phyletic size decrease is also clearly indicated by the different values of . The variation of is actually so wide that the coefficient of variation (V) ranges from 15 to 30 in these selected samples. In the two Pliocene samples Sh 1 and Nj (and probably also Sh 2), growth rings are, if present, rather weak, notwithstanding the fact that the shell often attains a size of over 30 mm. It is not known, however, whether this phenomenon too is attributable to phyletic change or simply due to different environmental conditions. Two other Late Pliocene samples, Ik and Kg 1, and all the Pleistocene samples contain a number of large individuals with clear stepwise growth rings. These fossil samples, except for the youngest one, Sm, generally have much larger values of than Recent samples. Therefore, not only the ultimate size but also the annual growth rate must have decreased with time. Though the correlation between the absolute age and loge is not very high (N=10, r=0.659, 0.01<P<0.05), the calculated slope of the regression line by the least square method indicates the evolutionary rate of this size decrease to be approximately 120 millidarwins.

The Late Pliocene sample Kg 1 shows unusually larger values of and than other Pliocene and Pleistocene samples. Individuals over 40 mm are commonly found in this sample. As is discussed and described in later pages, however, some taxonomic distinction may be required for this sample in view of the unusual mode of fossil occurrence and some other unique morphological characteristics.

2. Shape of Shell [see Tables 4-10]

The allometric equations of average relative growth between L and H, L and T, and L and D, as well as standardized form ratios H/L, T/L and D/L, were obtained on several fossil samples. The values of growth ratio (a) are sometimes significantly different between samples of different ages, but there is nothing like a definite trend.

The Pliocene sample Ik seems to show unusually smaller values of H/L (i. e., comparatively low outline) in every growth stage than other contemporary and younger samples (Plate 7, Figs. 9, 10). In contrast, the Middle Pleistocene sample Tm 1 (and also Tm 2) has decidedly larger values (i. e., comparatively tall outline) in comparison with other samples (Plate 7, Figs. 7, 8). These peculiar features are so conspicuous that one can perceive their uniqueness at a glance. However, no definite trend with time can be detected for this character and other form ratios.

3. Number of Radial Ribs [see Table 11]

The number of radial ribs on the disk (X), particularly its average and variability, is an easily recognizable and probably meaningful character on which chronological change can be discussed. This character was examined on five large Pliocene and seven large Pleistocene samples. These fossil samples show similarly unimodal and nearly normal frequency distribution in the number of radial ribs and comparable coefficient of variation (V) with Recent samples. Taking the aforementioned data on Recent samples into consideration, it is concluded that the average number of radial ribs has decreased considerably with time. The obtained values of in all the examined Pliocene samples (Sh 1, Nj, Ik, Kg 1 and Mz 2), which sometimes exceed 16.0, are evidently larger than those of the Pleistocene and later samples. The Early and Middle Pleistocene samples (Ob, Tm 1, Sn 3) seem to be intermediate in this character between Pliocene and Holocene-Recent ones. The values of of the Late Pleistocene samples (Ab 1, Jz, Sm), which are all from the Pacific coast of central Honshu, are within the range of the geographic variation in the present seas, but usually significantly larger in comparison with Recent samples of the same region.

As illustrated in Fig. 22, the obtained values of in these samples and their estimated absolute ages show a significant correlation (N=18, r=0.873, P<0.01). Among the Late Pliocene samples, Kg 1 (and also Kg 2) has a somewhat unusually large value, and is taxonomically separable from other samples in view of some other morphological and paleoecological characteristics. If this sample is excluded, the correlation coefficient becomes still higher. This is therefore regarded as an example of a directional change of phyletic evolution which took place during geological ages.

The geographic variation of is, as already described, considerably wide in Recent populations. Judging from the historical change of this character, the living populations on the Pacific side of central Honshu may be regarded as more advanced than those in the East China Sea and the Japan Sea. All the examined samples from the latter sea areas (EC 2, Ta 4, Hm) are comparable in this character with the Middle Pleistocene fossil samples (Tm 1, Sn 3) of the former area.

The oldest sample of C. vesiculosus, Sh 1, shows a value of somewhat smaller than other Pliocene samples. Though the fossil records of this stock become very obscure before the Late Pliocene, this sample may suggest that phyletic change was not necessarily unidirectional in Pliocene times.

4. Surface Sculpture (Phenotypic Frequency) [Table 12]

Discrimination between the two phenotypes Q and R is also possible in every fossil sample, even those in poorly preserved condition. Therefore, the phenotypic frequency N_R/N is a clean-cut and useful character for the study of phyletic change at the population level.

The historical change of this character was discussed in some detail in my previous paper (Hayami, 1973). My conclusion there about the gradual change of phenotypic frequency as well as the concept of phenotypic substitution as a cause of morphological break without any intermediate form (Hayami and Ozawa, 1975) is fundamentally supported by subsequently accumulated data.

Many fossil samples of C. vesiculosus were examined, but no individual belonging to the Phenotype R could be found in Pliocene and Early Pleistocene samples. The Middle Pleistocene samples Ij 1 and Ij 2 are also entirely composed of individuals belonging to the Phenotype Q. The earliest occurrence of the Phenotype R so far known on the Pacific side is from the Middle Pleistocene Sanuki Formation (samples Sn 1, 2, 3, all approximately 0.5 Ma). Though the observed phenotypic frequency is somewhat variable among these Sanuld samples, the largest sample, Sn 3, shows a value of N_R/N as high as 0.23. I have found a few individuals of the Phenotype R in a small sample (Dr. Ogasawara's private collection) from the Middle Pleistocene Shibikawa Formation of Oga Peninsula, signifying the nearly coeval appearance of this phenotype in populations on the Japan Sea side.

In the Late Pleistocene, a temporal transgression with strong influence of warm current occurred, leaving prolific fossil beds of "Jizôdô fauna" in the middle part of the Boso Peninsula. The samples Ny 1, 2, 3, 4, Jz, and Ab 1, 2 are attributable to this horizon (approximately 0.37 Ma). These samples often show somewhat lower phenotypic frequency than the Middle Pleistocene Sanuki samples, but the value of N_R/N is also considerably variable within the Jizôdô Formation. For example, the two large samples Ab 1 and Ab 2, which were randomly collected from two different layers about 2 meters apart of one and the same outcrop at Atebi, show significantly different phenotypic frequency; the sample Ab 1, taken from the lower layer, is richer in the Phenotype R than the sample Ab 2, taken from the upper layer. Within this brief period, therefore, minor fluctuation of phenotypic frequency may have occurred. The observed values of N_R/N in the large samples of this horizon are, however, confined to the range from 0.12 to 0.20, and it is probably appropriate to roughly estimate the average phenotypic frequency at the Jizôdô transgression is at 0.15.

The horizon of the sample Sm was actually determined by Sugihara et al. (1978), using marker tephras to be the upper part of the Yabu Formation (approximately 0.29 Ma). The value of N_R/N of this sample is not much different from the average value of the Jizôdô samples.

The geographic variation of the phenotypic frequency in Recent samples is, as stated before, considerably wide. However, large Recent samples from the Pacific coast of central Honshu seem to be fairly stable in this character; the observed frequency ranging from 0.39 to 0.46 is much higher in comparison with these Pleistocene fossil samples. The phenotypic frequency in the Recent samples from the East China Sea and Japan Sea is as low as from 0.17 to 0.30, but still seems to be higher than that of the Pleistocene ones. Because there is no large Middle or Late Pleistocene sample in the areas facing the East China Sea and Japan Sea, almost nothing has been known as to the geographic variation nf this character in the past. As shown in the collective illustration in Fig. 23, how ever, the chronological change of the phenotypic frequency may be said to have a trend, if not truly unidirectional.

The present conclusion is unexpectedly in harmony with the aforementioned phyletic decrease in the average number of radial ribs. This may not mean pleiotrophic effect of the same genetic factor, but it is not surprising to see that the phyletic change of the two independent characters, and probably also the related gene frequency, are the most advanced in the peripheral area of distribution (Pacific coast of central Honshu). The populations in the East China Sea and Japan Sea may be more conservative, the genetic nature having perhaps changed somewhat behind the Pacific side populations.

In 1973, I estimated the values of selection coefficient (s) as well as those of the change of gene frequency per one generation (Δq) based on an assumption of the simplest relation between genotypes and phenotypes. The genetic background of this dimorphism is still unknown, but, on the same assumption, the estimated values of s and Δq should be revised, because recent geochronology reveals that the absolute age of the Jizôdô Formation was greatly underestimated in my 1973 article. Assuming the same time span of one generation (two years on average), the following revised values are obtained:

1. In the case of dominant mutant gene [R-type individuals to be dominant homozygotes plus heterozygotes]
   s=8.6×10^-6
   Δq=5.7×10^-7 for fossil population of Jizôdô Formation
   Δq=1.2×10^-6 for Recent population of Sagami Bay
2. In the case of recessive mutant gene [R-type individuals to be recessive homozygotes only]
   s=1.2×10^-5
   Δq=1.1×10^-6 for fossil population of Jizôdô Formation
   Δq=1.8×10^-6 for Recent population of Sagami Bay

The result of the above calculations indicates for every case that a slight difference of adaptive value (only of the order of 10-5 on the average) between the individuals of the two phenotypes would explain the phyletic change of phenotypic frequency. Because their morphological difference is quite significant, I still maintain that natural selection is a probable cause for the observed phenotypic substitution. However, a slight difference of adaptability like this, if present, would not be practically testable with capture-release experiments in the field, because a sample composed of more than a million living individuals would be required for this test.

At the same time I would by no means deny the possibility of non-adaptive causes for this evolutionary change. For example, this slow rate of substitution may also be explicable by spontaneous mutation pressure (only of the order of 10-6 per one generation). The effect of random genetic drift, which is most certainly dependent on population size, is also a possible cause, for it is questionable whether such weak selection pressure or mutation pressure could concur with this effect. Regrettably, it must be concluded that the cause of this phenotypic substitution is almost beyond resolution. The current debate between the selectionists and the neutralists would still remain unresolved, even if more ideal fossil records were investigated.

Raup (1977), citing Hayami and Ozawa's (1975) data on the historical change of phenotypic frequency in the lineage of this pectinid (the same data as in my 1973 article), is of the opinion that our observed trend is not necessarily attributable to a deterministic cause like natural selection, because the null hypothesis of random walk cannot be rejected with 95 percent confidence. I agree with him that the trend may be caused by various nonadaptive factors, including stochastic ones. However, Feller's theorem, which was applied by Raup (1977) for the test of random walk, takes into account only the differential frequency of plus and minus signs in a time series of morphological change, and various other valid quantitative data such as net amount of change, number of contemporary samples, sample size, standard errors, and time span between samples are not utilized. I dare say, therefore, that the effectiveness of this test for this purpose is rather weak, and rpliable evaluation of random walt cannot necessarilybeachieved.