Graph 1Length measurement values of 511 shells of C. fimbriata. With its high total  X 2 (Chi-squared) error value, the plot cannot be considered “normal.” The threshold value for acceptance at the 0.05 probability level is ~36.

During a study of a very large lot of Hawaiian-collected Cypraea controversa Gray, I noted an odd relationship between one of its four aspect ratios and its labial tooth counts. Because this mid-Pacific variety of C. isabella Linné is rather distinct from populations in other Indo-Pacific areas, I turned my attention to a different species of cowry with a similiarly wide range. Cypraea fimbriata also occurs in Hawaii and seems to retain most of its characteristics, thus I was hoping to “see” what a relatively ‘normal’ (read “Gaussian”) cowry looks like. The subsequent search revealed—as it often may in ‘frontier science’—more about this animal, its shell and its local history than was expected.

Figure 1Typical live-collected Cypraea fimbriata; Len= 0.435", Wid= 0.244", Hgt= 0.188", Lab. th.= 19, Col. th.= 20

To start with, “normal” is not the word for the length plot of C. fimbriata (Graph 1). Rather, this plot shows a disturbed set of values that cannot come close to passing a test for being a Gaussian distribution. In an attempt to understand what created this bizarre shape, I again went into my database of these shells (collected from the upper Waianae coast of Oahu, Hawaii) and separated the two primary groups which comprise the lot. I then plotted their length values in Graph 2, using the same X-axis as Graph 1. (It spans from -3 to +3¼ standard deviations and has a mean of 0.414 inches.) The results offered food for thought!

Graph 2An “X-ray” of the green plot in Graph 1 reveals the two groups, live- and dead-collected shells.

Figure 2Typical dead-collected Cyp­raea fimbriata; Len= 0.392", Wid= 0.221", Hgt= 0.176", Lab. th.= 17, Col. th.= 18

There are about half as many dead-collected shells as there are live-collected. The plot of the larger of the two groups retains the overall shape of the original curve but the big surprise is in the remaining curve. It represents the dead-collected shells and these are almost certain to be several decades older than the larger, ‘fresh’ lot. As such, they should show what the population looked like earlier in the 20^th century (i.e., surely prior to Hurricane Iwa in 1982, but very likely pre-dating World War II). In Figures 1 & 2 are representatives of these two groups. Each shell has both traits (length & labial tooth count) that are the peak (mode) values for its group in Figures 2 & 4. (Click on a Figure for its original-sized image. Use the Back function of your browser to return to this page.)

As earlier studies had shown me that there is a strong correlation between the average length of cowries and their number of labial teeth, I used a common spreadsheet function to produce a correlation coefficient for the lengths and labial tooth counts of C. fimbriata. The value of +0.545 showed a strong—though not perfect—correlation. This was no surprise, but a few moments of thought brought up the question, “If the Length and Labial Teeth are so closely linked, would a plot of the distribution of labial tooth counts show corresponding irregularities?”

Graph 3The Gaussian curve created by the 511 shells in this lot. There is no obvious indication that two other “bell” curves make up this one.

Expecting to see disturbances in this new graph, I worked up the plot shown in Graph 3. This was a surprise,... but I had forgotten an item posted on this site more than a year ago showed that the labial tooth counts of C. fimbriata indeed had a “normal,” Gaussian-looking plot. Here, applying a X ² test for ‘fit’ yielded a value that was within the limit for acceptance as a “bell” (Gaussian) curve. That was all well and good, but this result was also unexpected. What were the influences causing the distorted length plot for the 511 shells of Cypraea fimbriata?

An old saying goes, “When all you have is a hammer,... everything looks like a nail.” So I did the same thing to this plot (Graph 3) as was done to Graph 1; I split out the dead- and live-collected shells from the database and plotted them on the same X-axis as Graph 3 to create Graph 4. There were the causes of the disheveled lengths plot! With a strong relationship between shell lengths and labial tooth counts, these two groups were the forces pushing up the maxima in Graph 1. And, yet,...

Graph 4Another “X-ray view.” This time it shows the two groups making the composite plot of labial tooth count values of the green line in Graph 3.

Something very curious was being brought to light. Graph 2 shows that the lengths of the current population are closely allied with those of the of all 511 shells and Graph 4 shows that the “living” population has very few of the lowest tooth counts, while Graphs 2 & 4 show that the dead-collected shells plainly are shorter in length and lack higher tooth counts (very few above 20). The Graph 2 curve says that the genetics that made the shells of the dead-collected groups (broken line) is slightly different from the genetics creating the current shells, with their increased labial tooth counts and, thus, longer shells.

How should one interpret these clues? If, on the one hand, Hurricane Iwa wiped out the original population except for a few survivors, how did the subsequent, re-established population suddenly develop increased tooth counts where it showed none before? However, if a “new,” longer group of Cypraea fimbriata was brought into Hawaiian waters by the heavy maritime traffic of World War II (or later arrivals), it should appear very much as the data show above—an additional pattern added to the original (older) variety of this cowry. Unfortunately, one (this one, at least) cannot account for the loss of the lowest labial tooth counts. Where have they gone?

To wrap this item up, I will add that there are more curious facts which have come to light while rummaging around all of the data and their relationships to one another: in addition to the overall labial tooth count curve meeting the acceptance level for a normal distribution, both the dead- and live-collected groups do so likewise, as separate populations; the average length of the dead-collected group is 0.393 inches while the current, living group is 0.425 inches; the average labial tooth count for the older shells is ~17¾ while the current population is ~19¼, the overall average is ~18¾.

There are more findings to pursue. This item is merely a ‘sidetrack’ from a ‘tangent’ to my study of Cypraea controversa's length-to-height / labial-tooth-count research. With one foot on the soapbox,... I maintain the deepest respect for the many works in conchology which have gone before, but the time has come for another round of  “standing on the shoulders of Giants.” I hope that the ancients will pardon if their toes get stepped on in attempts to get a clearer view. The 21^st century is here and, with it, the tools to “crunch” such vast amounts of data that our forebearers could not have imagined it in their wildest dreams.

Gotta go,... More things to do...

fin.