Alaska Department of Fish and Game produced the Kodiak Island king crab survey data. The underlying purpose of the survey was to gather biological and abundance information to be used in the management of king crab (opening and closing of fisheries, setting of quotas, etc.) It also presented a set of general questions and issues to stimulate the analysis. Major ones are as follows :
Originally the data set was intended to be analyzed statistically. Before starting the statistical analysis, an interactive visualization tool can help us to set a good direction to further analysis. I selected Spotfire as a visualization tool since it seemed to be suitable for visualizing a simple tabular data set like this King Crab data set. As a consequence, I found some interesting patterns in the data set, and they could be good pilot results for the actual statistical analysis.
The crab data set consists of 9 survey data files. Each of them has a different dimension, size, and category as follows.
| file | lines | columns | bytes | contentes |
| survey | 3450 | 14 | 244950 | the basic survey data |
| kodiak | 2687 | 2 | 48264 | map coordinates for the Kodiak Island shoreline |
| dstns | 1845 | 5 | 40590 | distributions of crab by size, year, category |
| fleet | 23 | 6 | 946 | commercial fleet/catch/price data, by year |
| catch | 96 | 4 | 2112 | commercial catch data, by year, district |
| eggs | 14 | 2 | 140 | average number of eggs per female, by year |
| salinity | 60 | 3 | 718 | ocean salinity, by month |
| celsius | 68 | 3 | 741 | ocean temperature, by month, in degrees celsius |
| fullness | 1170 | 7 | 37440 | distributions of females by clutch fullness |
Basically I tried to find rough answers to the questions described in the introduction section. So, I will present my visualization results that could give an idea about each question.
As you can find in the following scatter plots, the number of crabs was approximately decreasing through all the age levels and sexes. An interesting fact is that the number of eggs per female increased rapidly. Therefore we can conclude that the hatch rate decreased or the death rate of embryos was increased.
The following graph shows the number of vessels registered for fishing(red line), the number of crab caught(blue line), and the wholesale price of king crab in dollars per pound(black line). As you can see in the graph, the number of crab caught decreased even though the number of vessels registered increased. Therefore I can conclude that overfishing is not a major cause. However overfishing could be a minor cause which accellerates the decline.
- Do environmental factors contribute?
King Crab's spawning takes place in shallow water in late winter and early spring. Adult King Crab prefers temperatures of 0° to 5.5° C. Considering these two facts, the following results suggest that the environmental factors contributed to the declination of King Crab's abundance. I think that the low salinity level and high temperature prevented embryos from hatching. My reasoning is as follows.
Global warming effect might be a possible cause of the increase of celsius. And the high temperature makes the iceberg in artic regions melt into water, which results in the low salinity.
- Are we seeing natural population cycles?
The survey data covers at most 2 decades, so I cannot see any natural population cycle only from the data set since the life span of King Crab is almost 20 years.
The next figure is an animated gif file that shows the length distribution of adult female King Crab. We can see the gradual slight left shift of the peak as time goes by. It means that the King Crab society was getting older since the length of the King Crab generally is proportional to the age.
There were no special changes in the spacial distribution except that the population sizes decreased in all districts.
I could see a peak near 1976 in the graph titled "Average # of pre-recruit-4 male", and the peak gradually shift to the right in the following three graphs. The peak means that there are cohorts in the population at the time period since the crab population usually is known to be large when there are strong cohorts.
The aging process and length distribution of adult male are similar to those of adult female as depicted in the animated graph in question 3. This fact says something about the social interaction of crab with regard to sexes and age groups. There seems to be natural population balances between two sexes. I'm not sure but there are some interactions between male and female groups of similar size(or age).
As you can see in the graph titled "Number of eggs per female", there seems to be no problem in the fertility of females. As I mentioned already, however, the abundance of females seems to be in jeopardy. The small population of juvenile female also could be a potential jeopardy.
I tried to find some correlations by comparing the two graphs titled "Average # of adult female in a pot" and "Average # of recruit males", but I could not find any correlations between the two factors.
Spotfire is a powerful visualization tool. We can use Spotfire to get a brief but important idea about the raw data before we start more complex and difficult data analysis. Direct manipulation combined with interactive display makes it easy to find some important patterns in a relatively large raw data. Spotfire is effective to show some correlations among 2 or 3 data attributes. It is easy to learn, and the learning can be retained overtime.
However, Spotfire has some limitations. It does not support many commonly used aggregation functions(count, summation, average, standard deviation, etc.). I could not standardize the raw data to compare them in the same scale, so I should preprocess the raw data using MS Excel before importing them. Another limitation is that it difficult to visualize the hierarchical data and show the correlations in high-dimensional data set using Spotfire in a natural way.
